Sixth Hutton Symposium on the origin of granites and related rocks: proceedings of a symposium held in Stellenbosch, South Africa, 2-6 July 2007 (GSA Special Paper 472)

0 • ...I :g Sixth Hutton Symposium on The Origin of Granites and Related Rocks Proceedings of a Symposium held...

59 downloads 893 Views 21MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

..

<>

0

•

...I :g

Sixth Hutton Symposium on The Origin of Granites and Related Rocks Proceedings of a Symposium held in Stellenbosch, South Africa 2–6 July 2007

Guest Editor-in-Chief: John D. Clemens Department of Earth Sciences University of Stellenbosch Private Bag X1 7602 Matieland South Africa

EESTRSE Editor: Colin Donaldson Guest Editors: Carol D. Frost Alexander F.M. Kisters Jean-François Moyen Tracy Rushmer Gary Stevens

Copublished in volume format by arrangement with, and with permission of, The Royal Society of Edinburgh

Special Paper 472 3300 Penrose Place, P.O. Box 9140

Boulder, Colorado 80301-9140 USA

2010

© 2010 The Royal Society of Edinburgh COPYRIGHT: It is the policy of The Royal Society of Edinburgh not to charge any royalty for the production of a single copy of any one article made for private study or research. Specific permission will not be required for photocopying multiple copies of copyrighted material to be used for bone fide educational purposes, provided this is done by a member of the staff of the university, school, or other comparable institution, for distribution without profit to student members of that institution and provided the copies are made from the original publication. Requests for copying or reprinting of any article for any other purpose should be sent to the Royal Society of Edinburgh. The papers in this volume were originally published together as Earth and Environmental Science Transactions of the Royal Society of Edinburgh, Volume 100, parts 1 and 2 (ISBN 978 0 902198 31 9). Copies of that issue are available from the Royal Society of London’s distribution agents. Copyright is not claimed on any material prepared wholly by government employees within the scope of their employment. Copublished as a limited edition in volume format in the United States by The Geological Society of America, Inc., 3300 Penrose Place, P.O. Box 9140, Boulder, Colorado 80301-9140, USA. www.geosociety.org Printed in USA Library of Congress Cataloging-in-Publication Data Hutton Symposium on the Origin of Granites and Related Rocks (6th : 2007 : Stellenbosch, South Africa) Sixth Hutton Symposium on the Origin of Granites and Related Rocks : proceedings of a symposium held in Stellenbosch, South Africa, 2-6 July 2007 / guest editor-in-chief, John D. Clemens. p. cm. -- (Special paper ; 472) Includes bibliographical references. ISBN 978-0-8137-2472-0 (pbk.) 1. Granite--Congresses. I. Clemens, John D. II. Title. QE462.G7H88 2007 552’.3--dc22 2010027738 Cover: Photomicrograph of a Tasmanian dolerite containing residual silicic interstitial glass with a major element composition very similar to associated granophyric sills. Crossed polars; width = 4 mm. (From paper by S. Turner and T. Rushmer, “Similarities between mantle-derived A-type granites and voluminous rhyolites in continental flood basalt provinces.”)

10 9 8 7 6 5 4 3 2 1

ii

CONTENTS J. D. CLEMENS Preface

v

Sharad MASTER Plutonism versus Neptunism at the southern tip of Africa: the debate on the origin of granites at the Cape, 1776–1844

1

Herve´ MARTIN, Jean-Franc¸ois MOYEN and Robert RAPP The sanukitoid series: magmatism at the Archaean–Proterozoic transition

15

J.-F. MOYEN, D. CHAMPION and R. H. SMITHIES The geochemistry of Archaean plagioclase-rich granites as a marker of source enrichment and depth of melting

35

Simon TURNER and Tracy RUSHMER Similarities between mantle-derived A-type granites and voluminous rhyolites in continental ﬂood basalt provinces

51

Luke LONGRIDGE, Roger L. GIBSON and Paul A. M. NEX Structural controls on melt segregation and migration related to the formation of the diapiric Schwerin Fold in the contact aureole of the Bushveld Complex, South Africa

61

Jean H. BEDARD Parental magmas of Grenville Province massif-type anorthosites, and conjectures about why massif anorthosites are restricted to the Proterozoic

77

Paul D. BONS, Jens K. BECKER, Marlina A. ELBURG and Kristjan URTSON Granite formation: Stepwise accumulation of melt or connected networks?

105

Eric HORSMAN, Sven MORGAN, Michel de SAINT-BLANQUAT, Guillaume HABERT, Andrew NUGENT, Robert A. HUNTER and Basil TIKOFF Emplacement and assembly of shallow intrusions from multiple magma pulses, Henry Mountains, Utah

117

M. O. M. RAZANATSEHENO, A. NEDELEC, M. RAKOTONDRAZAFY, J. G. MEERT and B. RALISON Four-stage building of the Cambrian Carion pluton (Madagascar)

133

Keith BENN Anisotropy of magnetic susceptibility fabrics in syntectonic plutons as tectonic strain markers: the example of the Canso pluton, Meguma Terrane, Nova Scotia

147

J. D. CLEMENS, P. A. HELPS and G. STEVENS Chemical structure in granitic magmas – a signal from the source?

159

R. C. ECONOMOS, V. MEMETI, S. R. PATERSON, J. S. MILLER, S. ERDMANN and J. Z {A uK Causes of compositional diversity in a lobe of the Half Dome granodiorite, Tuolumne Batholith, Central Sierra Nevada, California

173

Axel MU } LLER, Alfons M. van den KERKHOF, Hans-Ju¨rgen BEHR, Andreas KRONZ and Monika KOCH-MU } LLER The evolution of late-Hercynian granites and rhyolites documented by quartz – a review

185

C. JUNG, S. JUNG, E. HELLEBRAND and E. HOFFER Trace element constraints on mid-crustal partial melting processes – A garnet ionprobe study from polyphase migmatites (Damara orogen, Namibia)

205

M. P. SEARLE, J. M. COTTLE, M. J. STREULE and D. J. WATERS Crustal melt granites and migmatites along the Himalaya: melt source, segregation, transport and granite emplacement mechanisms

219

Takashi HOSHIDE and Masaaki OBATA Zoning and resorption of plagioclase in a layered gabbro, as a petrographic indicator of magmatic diﬀerentiation

235

AUTHOR INDEX

251

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, v–vi, 2010 (for 2009)

Preface J. D. Clemens Department of Earth Sciences, University of Stellenbosch, Private Bag X1, 7602 Matieland, South Africa The Sixth Hutton Symposium on the Origin of Granites and Related Rocks was held on July 2–6, 2007 at the University of Stellenbosch, South Africa, founded on granite, nestled at the feet of towering mountains and fringed by the rolling winelands of the Western Cape. This Special Issue opens with Master’s historical account of how the Cape granites inﬂuenced 18th and early 19th century thinking on the origins of these rocks. The fascinating fact is that the granites of the Western Cape were apparently the ﬁrst intrusive granites recognised outside Britain. The balance of the volume contains a collection of research papers derived from the meeting and illustrates some of the important directions in which granite research may be evolving. One of the characteristics of the papers and talks presented at the meeting was that there seemed to be some shift in interest, away from the crust as a source of granitic magmas and towards mantle rocks that have been metasomatised by subduction-zone ﬂuids or melts. Nevertheless, the crust still holds pride of place as the cradle of granite genesis. The next 15 papers fall into four groups. Those concerning the origins of the magmas themselves include Martin et al., Moyen et al., Turner & Rushmer, Longridge et al. and Be´dard. Martin and his co-workers deal with sanukitoids and Closepettype magmatism. Both these kinds of magmas are eﬀectively unique to the period of time that marks the transition between the Archaean and the Proterozoic, and both are believed to be produced by partial melting of enriched mantle, rather than the crust. Martin et al. present their ideas on how the Archaean mantle was enriched, and they correlate the production of these distinctive magmas with the temporal evolution of heat production in the planet. The enrichment theme is continued in the paper by Moyen et al., who describe geochemical diﬀerences among Archaean TTG rocks (tonalites, trondhjemites and granodiorites) that lead them to recognise three subgroups whose chemistry they interpret as reﬂecting the degree of source enrichment and the depth of melting. Turner & Rushmer draw a parallel between the continental ﬂood basalts and some A-type granites that they interpret as being mantle-derived. They note the geochemical similarities between rhyolitic rocks that cap ﬂood basalt sequences and some kinds of A-type granites. They suggest that these A-type granitic rocks were produced by fractionation of basaltic parent magmas. Staying with the theme of maﬁc magmas, Longridge et al. examine the thermal eﬀects of the Bushveld Complex on underlying metapelites. They show how partial melt from the metapelites was segregated, and they describe structures that indicate slow, buoyant diapiric ascent of the felsic magma, attended by ductility enhancement that allowed the formation of marginal shear zones. Be´dard presents a model for the production of diﬀerent types of anorthositic magmas by partial melting either of basaltic arc crust or of mantle enriched by subductionderived ﬂuids. He goes on to speculate as to why anorthositic magmas may or may not be present in a terrane, depending on temporal changes in lithospheric heat-production. The papers dealing with the construction of plutons and emplacement include Bons et al., Horsman et al., Razanatseheno

et al. and Benn. Considering the mechanisms of melt transfer from partially molten source to pluton, Bons et al. challenge the idea of a continuum model, with small veins progressively feeding ever-larger veins and dykes. Instead, they outline the features of, and evidence for, a model involving stepwise accumulation of melt volumes, arguing that the full range of vein/dyke sizes never coexists in nature. Horsman et al. describe evidence for the multi-pulse assembly of some shallowlevel plutons in a tectonically-quiescent regime. They describe a progression from sill to laccolith to piston-type emplacement as the magma systems increase in volume. Focusing on a crudelyzoned pluton in Madagascar, Razanatseheno et al. provide structural and AMS (anisotropy of magnetic susceptibility) evidence of steeply-inclined magmatic foliations and lineations reﬂecting upward ﬂow of the magmas, to ﬁll the pluton in pulses of diﬀerent composition. Benn continues the theme of AMS as a tool to study the structural evolution of plutons during and after their magmatic histories. He demonstrates that fabrics formed during the emplacement of syn-tectonic plutons can be distinguished from the eﬀects of postemplacement deformation, and that the post-emplacement fabrics can be used to shed light on regional bulk kinematics. Several of the papers are concerned with the mechanisms by which granitic plutons acquired their internal compositional diversity. Clemens et al. review the evidence from a broad range of granitic bodies and ﬁnd a common lack of evidence for signiﬁcant degrees of diﬀerentiation. Departing from conventional wisdom, they conclude that the heterogeneities in most granitic bodies were not brought about through crystal fractionation, magma mixing, restite unmixing or any other diﬀerentiation mechanism, but were inherited from the magma sources and only slightly modiﬁed thereafter. In contrast, Economos et al. examine the causes of compositional diversity in the Tuolumne batholith, taking the Half Dome granodiorite as their example, and conclude that fractionation was the dominant process. Taking a less conventional approach to these kinds of problems, Mu¨ller et al. review the evolution of Late Hercynian felsic magmas through the study of quartz crystals in the rocks. They use cathodoluminescence imaging, Fourier-transform infrared spectroscopy and electron probe microanalysis to reveal the growth and alteration features that reﬂect the changes in magma compositions (including dissolved H2O content) and crystallisation conditions. Features such as adiabatic and non-adiabatic magma ascent, temporary storage of magma and mixing with more maﬁc magma can all be discerned, illustrating the utility of this technique for tracing a magma’s history. The balance of the volume consists of three papers presenting speciﬁc case studies that range from work on prospective protoliths for granitic magmas and the processes of melting and magma migration to a study of the genesis of some anorthositic magmas. Jung et al. use garnet REE zoning patterns to reveal the intricacies of chemical equilibrium and disequilibrium during partial melting of metasedimentary rocks in the Damara orogen of Namibia. The results have importance for the understanding of REE patterns in

2009 The Royal Society of Edinburgh. doi:10.1017/S175569100901620X

vi

PREFACE

crustally-derived granitic magmas. In the context of granitic magmatism in continental collision zones, Searle et al. investigate the formation of Himalayan granites. They ﬁnd that, here, ﬂuid-absent, mid-crustal melting of the Proterozoic protolith was due to a combination of thermal relaxation and high rates of internal heat production. Both mantle heat sources and the oft-cited shear-heating mechanism are ruled out. Finally, in a return to the mantle whence, it has been claimed, all good things come, Hoshide & Obata describe the fractionation mechanisms, in the Murotomisaki gabbro (Japan), that led to the production of layers of felsic magma. Surprisingly, it seems that ﬂushing by aqueous ﬂuids led to preferential remelting of plagioclase, and the formation of anorthositic layers that spawned diapiric upwellings within the magma body.

From the above it will be apparent that there is a great degree of variety here, and something for nearly every taste. If these papers represent the present thrust of granite-related research, it would seem that we can expect future progress particularly in understanding (1) the melting processes that lead to the formation of granitic magmas; (2) the physical, chemical and kinetic controls on the compositions of granitic magma; and (3) the relationships between the production of magma pulses and the mechanisms and time-scales of pluton construction and magma ﬂow within plutons. As with most predictions, this is probably incorrect. In any case, it will be fascinating to see what themes dominate in the next Hutton Symposium.

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 1–13, 2010 (for 2009)

Plutonism versus Neptunism at the southern tip of Africa: the debate on the origin of granites at the Cape, 1776–1844 Sharad Master Economic Geology Research Institute, School of Geosciences, University of the Witwatersrand, P. Bag 3, WITS 2050, Johannesburg, South Africa Email: [email protected] ABSTRACT: The Cape Granites are a granitic suite intruded into Neoproterozoic greywackes and slates, and unconformably overlain by early Palaeozoic Table Mountain Group orthoquartzites. They were ﬁrst recognised at Paarl in 1776 by Francis Masson, and by William Anderson and William Hamilton in 1778. Studies of the Cape Granites were central to some of the early debates between the Wernerian Neptunists (Robert Jameson and his former pupils) and the Huttonian Plutonists (John Playfair, Basil Hall, Charles Darwin), in the ﬁrst decades of the 19th Century, since it is at the foot of Table Mountain that the ﬁrst intrusive granites outside of Scotland were described by Hall in 1812. The Neptunists believed that all rocks, including granite and basalt, were precipitated from the primordial oceans, whereas the Plutonists believed in the intrusive origin of some igneous rocks, such as granite. In this paper, some of the early descriptions and debates concerning the Cape Granites are reviewed, and the history of the development of ideas on granites (as well as on contact metamorphism and sea level changes) at the Cape in the late 18th Century and early to mid 19th Century, during the emerging years of the discipline of geology, is presented for the ﬁrst time. KEY WORDS: Abel, Anderson, Barrow, Darwin, Degrandpre´, Hall, Hamilton, Hausmann, Itier, Jameson, Krauss, Masson, Paarl, South Africa, Table Mountain

The Cape Granites, situated in the Western Cape Province of South Africa, at the southern tip of the African continent, are a granitic suite dated at c. 550–510 Ma, intruded into Malmesbury Group greywackes and slates of the Neoproterozoic Saldanian Belt (Armstrong et al. 1998; Da Silva et al. 2000). These Neoproterozoic rocks are unconformably overlain by the early Palaeozoic Cape Supergroup, of which the lowermost Table Mountain Group, comprising mainly orthoquartzites, constitutes the top two-thirds of Table Mountain in Cape Town (Moore 1994; Compton 2004). Whilst aspects of the geology of Table Mountain and the Cape Granites have been studied for more than two centuries, it is not generally known that the Cape Granites have played a not insigniﬁcant role in the history of granite research. Actually, studies of the Cape Granites were central to some of the early debates between the Neptunists and the Plutonists, at the end of the 18th and in the ﬁrst decades of the 19th Century, yet they seem to have escaped the attention of standard histories of the subject. The Neptunists, led by the ideas of Abraham Gottlob Werner (1749–1817) (Fig. 1a) believed that all rocks, including granite and basalt, were precipitated from the primordial oceans (Werner 1787, 1791; see discussions in Adams 1938; Hallam 1983). The Plutonists, who followed the ideas of James Hutton (1726–1797) (Fig. 1b), believed in the intrusive origin of some igneous rocks, such as granite, which were believed to have been injected, in a molten state, into previously existing rocks (Hutton 1795; Adams 1938; Hallam 1983). In this paper, some of the early descriptions and debates concerning the Cape Granites are reviewed, and the history of the development of the ideas on granites at the Cape in the late 18th Century and early to mid 19th Century is presented for

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016193

the ﬁrst time. Extensive quotes are given from the original literature, much of which is in extremely rare books and obscure journals (which may explain why this chapter in the history of granite studies has not been better known).

1. Geology of Table Mountain and the Cape Granites – early accounts Ever since the discovery of the sea route to India via the Cape of Good Hope by Portuguese navigators in the late 15th Century, Table Bay has been a favoured stopping place for ships travelling between Europe and Asia. The ﬁrst known ascent of Table Mountain was by the Portuguese explorer Antonio de Saldanha in 1503 (Raven-Hart 1967). There are many detailed descriptions of the topography of Table Mountain by travellers and explorers in the 17th and 18th Centuries, e.g., by Sir Thomas Herbert (1634); Guy Tachard (1686); Peter Kolb (1719); Franc¸ois Valentyn (1726); the Abbe´ Nicolas Louis de la Caille (1763), who even named a constellation (Mensa) after the mountain; Carl Frederick Brink (1778); and Otto Mentzel (1785). The Cape Granites were ﬁrst recognised (as a variety of ‘saxum’ or granite), not in Cape Town, where they outcrop prominently, but at Paarl (Fig. 2), some 50 km to the NE, by the Scottish botanist Francis Masson (1741–1805) (Fig. 3). In 1772 Masson accompanied the distinguished Swedish botanist Carl Peter Thunberg on a journey to the interior of South Africa (Masson 1776, 1994; Thunberg 1788). Masson (1994) noted in his journal for 17th December 1772: ‘‘17th, I went up to the top of the Perel Berg, where I spent a whole day in search of plants, and hunting a sort

2

SHARAD MASTER

Figure 1 (a) Abraham Gottlob Werner (1749–1817), founder of the Neptunist School. Engraving by Ambroise Tardieu after an original portrait by Vogel. Source: Wikimedia Commons (b) James Hutton (1726–1797), founder of the Plutonist School. Image courtesy of the Scottish National Portrait Gallery: Sir Henry Raeburn, James Hutton, Scottish National Portrait Gallery; Purchased with the aid of The Art Fund and the National Heritage Memorial Fund 1986.

of antelope called Ree Bock; but had no success. I saw nothing here so worthy of observation as two large solid rocks, of a roundish ﬁgure; each of which, I may

Figure 2 (a) View of Paarlberg (‘Pearl Mountain’), above the town of Paarl, showing the two granite domes referred to today as the Suider Paarl (‘Southern Pearl’, left) and Noorder Paarl (‘Northern Pearl’, right). (b) View of bald granite monoliths making up the Paarl Mountain. Note the vertical crevasses described by Pierre Sonnerat (1782) and by Dugald Carmichael in 1806. (c) Aplite dykes cutting the Paarl Granite, described by Anderson (1778), and water-ﬁlled weathering basins, described by Sonnerat (1782).

positively say, is more than a mile about the base, and upward of two hundred feet high above the ground. Their surfaces are nearly smooth, without chink or ﬁssures, and they are found to be a species of saxum or

PLUTONISM VERSUS NEPTUNISM AT THE SOUTHERN TIP OF AFRICA

3

country, they being commonly divided, or composed of diﬀerent strata . . .’’ Anderson observed, however, a second variety of rock cutting across the main mass: ‘‘near its North end a stratum of a more compact stone runs across, which is not above twelve or fourteen inches thick, with its surface divided into little squares, or oblongs, disposed obliquely. This stratum is perpendicular; but whether it cuts the other to its base, or is superﬁcial, I cannot determine . . . I have sent a specimen of the rock and of the stratum, which are both what the mineralogists call saxa conglutinata or aggregata, and consequently are diﬀerent from the more solid stones which constitute the greatest part of the mountains here; and is likewise another proof of its being a single stone.’’

Figure 3 Francis Masson (1741–1805), Scottish botanist who ﬁrst discovered granite at Paarl Mountain. Detail from a painting by George Garrard. From Masson (1994).

granite, diﬀerent from that which compose the neighbouring mountains.’’ Masson did not have time to examine the ‘Perel Berg’ or Paarl (Pearl) mountain in more detail. However, he persuaded William Anderson, who had been surgeon’s mate and naturalist on Captain James Cook’s voyages, to make a more detailed study (Forbes 1965). Anderson visited the Paarl Mountain sometime in 1776, and made an attempt to determine its size and shape, and its geological constitution. In a letter dated 24th November, 1776, sent from Cape of Good Hope, Anderson communicated his ﬁndings to Sir John Pringle (then President of the Royal Society), who had been involved with Cook’s expedition. Pringle published the letter (which was read on 15th January, 1777) in the Philosophical Transactions of the Royal Society in 1778; meanwhile Anderson went on to join Cook’s third voyage to the Paciﬁc Ocean, where he perished. Anderson (1778) described the Paarl Mountain as: ‘‘a stone of extraordinary size . . . The Stone is so remarkable that it is called by the people here the Tower of Babel, and by some the Pearl Diamond. . . . It is of an oblong shape, and lies North and South. The South end is highest; the East and West sides are steep and high; but the top is rounded, and slopes away gradually to the North end, so that you can ascend it by that way, and enjoy a most extensive prospect of the whole country.’’ He estimated its circumference as ‘‘exceeding half a mile’’, and as for its height, he ventured to say ‘‘it equalled the dome of St Paul’s Church’’. As to its monolithic nature, Anderson (1778) observed that ‘‘it would certainly impress every beholder, at ﬁrst sight, with the idea of its being one stone, not only from its ﬁgure, but because it is really one solid uniform mass from top to bottom, without any interruption; which is contrary to the general character of the high hills of this

Note that the second, ‘more compact’ stratum is an aplite dyke, but was not recognised as such by Anderson (1778), since it was only some 17 years later that Hutton (1795) ﬁrst described intrusive dykes. The volcanologist Sir William Hamilton (1730–1803), famous for his descriptions of Vesuvius and the Eifel District of Germany, studied Anderson’s samples of the Paarl granites sent to him by Sir John Pringle. In a brief letter written at Grosvenor Place on 25th July 1777, and published as an appendix to Anderson’s article, Hamilton (1778) stated: ‘‘I return you many thanks for the sight of the stones from the Cape of Good Hope. I have not time to examine them very minutely; but they seem to be both of the same nature, granites, the smaller piece being only of a ﬁner texture. The highest points of the Alps are composed of granite of the same nature, and seem to have been lifted up by exhalations, volcanic explosions, or some such causes. This singular immense fragment of granite most probably has been raised in the same manner. Most of the mountains which are called primitive (which I believe is only a term) are of this texture.’’ The famous Paarl Mountain was also described by many subsequent travellers to the Cape. The French botanist Pierre Sonnerat (1748–1814), who journeyed to Asia between 1774 and 1781, and stopped at the Cape, called it the ‘Montagne de la Perle’, and described it as follows (Sonnerat 1782, II, 91; translated by SM): ‘‘The Paarl Mountain, which lies several leagues within the country, merits a description, being one of the highest in the vicinity of the Cape. It is composed of a single block of granite crevassed in many places. Nature has fashioned near the summit various caves and depressions, where one ﬁnds crystals of white and yellow rocks.’’ John Barrow (1801) mentioned the observations of the Paarl Mountain by Masson, Anderson and Hamilton in the Philosophical Transactions, and noted that from their descriptions, ‘‘it would appear that these two masses of stone rested upon their own bases, and were detached from the mountain; whereas they grow out, and form a part, of it.’’ He regarded the Paarl granite as being identical to the granite he had found at the foot of Table Mountain, and described it as being made of ‘‘aggregates of quartz and mica; the ﬁrst in large irregular masses, and the latter in black clumps resembling shorl [sic]: they contain also cubic pieces of feltspar [sic], and seem to be bound together by plates of a clayey iron stone.’’ Playfair

4

SHARAD MASTER

(1802, pp. 399, 410) in his popular explication of Hutton’s Theory of the Earth, concurred with the view that the Paarl Rocks were part of the mountain whose summit they form; however, he erred in attributing to Barrow the statement that the Paarl Rocks lie upon sandstone strata (Forbes 1965, p. 139). Barrow’s (1801) descriptions of the Paarl granites were further commented on by Louis O’Hier Degrandpre´ (1801b), who had translated Barrow’s work into French, and by the German naturalist Hinrich Lichtenstein (1811–1812). Captain Dugald Carmichael (1772–1827) was a Hebridean soldier, surgeon and naturalist knowledgeable in botany, geology and ichthyology (Smith 1831). He participated in the British occupation of the Cape in 1805–1806, but managed some geological observations during his military duties, noting in his journal for the 17th January 1806 (Smith 1831, p. 15): ‘‘Notwithstanding the fatigue of a nocturnal march, curiosity prompted me to walk up to the top of this hill, to which the colonists, struck by some peculiarity in its appearance, have given the name of Paarlberg. The summit is of granite, worn into a hemispherical form, and furrowed here and there by deep ﬁssures, through which the atmospherical moisture, condensed from the clouds, gushes down in perpetual rills. . . . On top of this granitic cupola, a number of detached masses of the same material lie scattered about, some of them apparently so nicely poised, that a slight push might roll them down upon the village.’’ The botanist Carl Peter Thunberg, who had accompanied Masson, published in 1788 a detailed account of his travels at the Cape. He had climbed Table Mountain no less than ﬁfteen times, and gave a basic description of its geological structure, as follows: ‘‘The uppermost strata are quite horizontal, but the lower ones lie in an oblique position. At the top, the rock appears to be a kind of sandstone, or volcanic ash; the middle stratum trapp, and the lowermost slate’’ (Thunberg 1986). The uppermost strata are made of quartzitic Table Mountain sandstone (no volcanics are present); the ‘trapp’ that makes up the middle stratum is most likely the Cape Granite; and the ‘slate’ (originally ‘skiﬀer’, Thunberg 1788, I, p. 251) of the lowermost strata refers to the greywackes and shales of the Malmesbury Group. Another Swedish botanist, Anders Sparrman, who was a friend of Thunberg and, like him, a prote´ge´ of Carolus Linnaeus, also visited the Cape in the late 18th Century, in 1772, and 1775–1776 (Forbes 1965), and wrote a valuable account of his travels, ﬁrst published in Swedish in 1783. Sparrman (1783) believed that the sandy wastes of the Cape Flats (which separated Table Bay from False Bay, and connected the Cape Peninsula with the Hottentots Holland Mountains near Stellenbosch) had formerly been covered by the sea. However, he appeared to believe that the exposure of these tracts was brought about not by a drop in sea level, but that they had been formed ‘‘particularly with sand, sea-shells, trunks of trees and such like rubbish’’ driven by the ‘‘violence of the south-east wind in False Bay’’ (Sparrman 1785; Forbes 1965, 1977). The French adventurer and ornithologist Francois le Vaillant (1795), believed, like Sparrman, that a fall in sea level at the Cape of Good Hope was evidenced by the sand dunes, sea shells and low elevations of the Cape Flats, which showed that they must have been recently submerged. He extended his hypothesis to include the interior mountains, far from the sea, which he believed had also been partly covered by the sea, and the retreat of this universal ocean was as postulated under the Neptunian hypothesis (Forbes 1965, pp. 126–127, 1977). Le

Vaillant (1795, I, p. 143) also remarked that the chain of mountains making up the Cape Peninsula, extending from Table Mountain to the Cape of Good Hope, was made up of granite. The geological structure of Table Mountain was ﬁrst described in detail by Louis Degrandpre´ (1801a) and Barrow (1801). Degrandpre´ had visited the Cape in 1793 (Kennedy 1954). He made observations concerning the geological constitution of Table Mountain (he thought that the summit of Table Mountain was made of granite), and he argued that the sea level had been much higher in the past, isolating the Cape Peninsula as an island. Degrandpre´’s (1801a) notions of the receding of the sea and the emergence of the Cape Flats, similar to those of Le Vaillant (1795) and Barrow (1801), were derived from the ideas of Delisle de Sales (1779). Delisle de Sales, nom de plume of Jean-Baptiste Claude Izouard, was the author of a 52-volume work, La Nouvelle Histoire des Hommes, which dealt with numerous subjects, including subterranean geography, the foundations of a new cosmogeny, volcanoes and earthquakes, famous discoveries of the primitive earth, etc., in which a three-volume subsection, Histoire du Monde Primitif, formed the main source of Degrandpre´’s (1801a) geological speculations concerning the Cape. Delisle de Sales (1779) was in turn, like the Neptunists, probably inﬂuenced by the speculative cosmogenies of Burnet (1691) and Buﬀon (1774). Barrow (1801), who had visited the Cape in 1797 and 1798, observed that the shore of Table Bay and the substratum on which Cape Town is built, is composed of ‘‘a bed of a blue, compact schistus’’. Upon the schistus, Barrow noted, ‘‘lies a body of strong pale yellow to deep red clay abounding with mica, which seems to have been formed from the decomposition of granite, immense blocks of which are embedded in the clay’’. Barrow further observed that ‘‘resting on the granite and clay is the ﬁrst horizontal stratum of the Table Mountain, commencing at about ﬁve hundred feet above the level of the sea’’. Barrow (1801) noted the presence of beds of sea shells, buried under vegetable earth and clay at a height of no less than three hundred feet (w100 m) above sea level. He commented that ‘‘the human mind can form no idea as to the measure of time required for the sea to have progressively retreated from such elevations’’. The geological observations of Barrow and Degrandpre´ were noted by the French traveller Jacques Ge´rard Milbert (1766–1840), who published an account of his travels to the Cape in 1812 (Milbert 1812; translated into German as Milbert 1825). Milbert (1812, 1825, p. 558) posed the question of what Table Mountain was made of, and simply gave, without taking sides, the answers of Barrow (1801), who thought that the summit was made of sandstone, and of his translator Degrandpre´, who had reiterated his original opinion that it was made of granite (Degrandpre´ 1801a) in his notes accompanying his translation of Barrow (Degrandpre´ 1801b). Milbert (1812) included a view of Table Mountain among the illustrations accompanying his account (Fig. 4a).

2. Plutonism versus Neptunism at the Cape Sir James Hall (1761–1832) was a chemist and geologist, and is regarded by some as the founder of experimental geology (e.g., Hall 1798; Craig & Jones 1985, pp. 162–163). Hall and John Playfair (1748–1819), both professors at the University of Edinburgh, had been friends and disciples of Dr James Hutton. Captain Basil Hall (1788–1844) was the second son of Sir James, and was educated at the High School of Edinburgh. He had a successful naval career, and wrote many articles and

PLUTONISM VERSUS NEPTUNISM AT THE SOUTHERN TIP OF AFRICA

5

Figure 4 (a) Sketch entitled: ‘Cape of Good Hope. View of Table Mountain’, from Milbert (1812). (b) Sketch showing the topography and geology of Table Mountain and surroundings, after information supplied by Dr Adam of Calcutta, from Jameson (1819).

multi-volume works about his naval life, ﬁrst for ‘‘young persons’’ (Hall 1831), reprinted later as a set of autobiographical sketches (Hall 1861). A biographical account of Basil Hall is given by Anonymous (2006). Basil Hall (1831, pp. 22–23) described how, as a youth, he met Professor Playfair at a house in the country, and how Playfair patiently explained to him the use of a sextant. He went on to say (Hall 1831, p. 179):

seeing Nature, as it were, with her face washed, more frequently than most other observers; and can seldom visit any coast, new or old, without having it in their power to bring oﬀ something interesting to inquirers in this branch of knowledge. That is, supposing they have eyes to see, and capacity to describe, what meets their observation.’’

‘‘About this period I began to dabble a little in geology, for which science I had acquired a taste by inheritance, and, in some degree, from companionship with more than one of the Scottish school, who, at the beginning of this century, were considered more than half-cracked, merely for supporting the igneous theory of Dr. Hutton, which, with certain limitations and extensions, and after thirty years of controversy, experiment, and observation, appears now pretty generally adopted. Sailors, indeed, have excellent opportunities of making geological observations, for they have the advantage of

In July 1812, Captain Basil Hall visited the Cape of Good Hope, putting into False Bay, and made an excursion to Table Mountain (Fig. 5), where he described the contact between the Cape granite and the Malmesbury greywacke, which he called ‘killas’, in letters to his father, Playfair and others. Playfair (Fig. 6a) published these observations (Playfair & Hall 1813) as a crucial example of the intrusive origin of granite, in support of Huttonian Plutonism, and in direct opposition to Wernerian Neptunism, represented at that time by his colleague at Edinburgh University, Robert Jameson (Fig. 6b), the arch-disciple of Werner in Scotland.

6

SHARAD MASTER

Figure 5 Map of the Cape Peninsula and False Bay, from Playfair & Hall (1813).

Captain Hall, in a letter to his father Sir James, recounted his visit to Table Mountain, along a ravine now known as Platteklip Gorge – the most popular and steepest route to the top (Playfair & Hall 1813): ‘‘I came after a short ascent, to a space where many yards of rock were laid perfectly bare, and I found myself walking on vertical Schistus, or on what might be called Killas. This rock was in beds highly inclined and stretching from east to west, which is nearly the direction of the mountain...On looking forward a little higher up, I saw another portion of rock that was also laid bare, and which appeared to be Granite. I had now no doubt of reaching in a few minutes the precise junction of the two rocks, and I ventured to predict to my companion, who was not a little surprised at the pleasure I seemed to feel on this occasion, that we should immediately see veins from the main body of the granite, penetrating into the rock on which we were now standing. In this I was not deceived; the contact was the ﬁnest thing of the kind I ever saw; the Windy Shoulder itself not excepted. The number of veins that we could distinctly trace to the main body of the granite was truly astonishing; and the ramiﬁcations, which extended on every side, were of all sizes, from the breadth of two yards to the hundredth of an inch. Masses of killas, cut oﬀ entirely from the main body of that rock, ﬂoated in the granite, without numbers, especially near the line of contact, and the strata appeared there broken, disordered, and twisted in a most remarkable degree. From this point, following up the course of the stream for about 300 yards, I found the whole a solid mass of granite. The granite is characterised by large crystals of feldspar, which, indeed, is true of all the granite which I met with at the Cape. Besides quartz and mica, large masses of hornblend [sic] enter occasionally into the composition of this rock. After ascending about 300 yards farther, I came to a line where the granite ceased, and was succeeded by strata of superincumbent Sandstone. These strata were horizontal, and without any symptom of disturbance or violence whatsoever. There was not a shift nor a vein; and this junction formed a most marked contrast with that which we had left below. Looking round from the point where I now stood, to all the parts of the amphitheatre, in the centre of which I was placed, I could trace the same line of junction, extending horizontally on every side.

Figure 6 (a) Professor John Playfair (1748–1819), the principal advocate of Huttonian Plutonism. Source: Wikimedia Commons (b) Professor Robert Jameson (1774–1854), principal advocate of Wernerian Neptunism. Source: Wikimedia Commons.

From this point, where the sandstone was ﬁrst discovered, for about 150 or 200 feet perpendicular, the rock continued of the same kind, viz. a red sandstone, in horizontal beds of no great thickness. From thence all the way to the summit the sandstone was of a much more indurated kind, quite white, and having pieces of water-worn quartz imbedded in it, from the size of a pea to that of a potatoe. The top is a plane of about ten acres, somewhat uneven, though, on the whole, nearly level.’’

PLUTONISM VERSUS NEPTUNISM AT THE SOUTHERN TIP OF AFRICA

7

Playfair, in a footnote, stated that the Windy Shoulder referred to by Hall ‘‘is on the side of Loch Ken in Kirkcudbrightshire, and is remarkable for veins of Granite, of the same kind with those here described’’ (Playfair & Hall 1813, p. 273). Commenting on Hall’s observations, Playfair remarked that ‘‘the phenomena here described point out two separate epochas [sic], distinguished by very diﬀerent conditions of the substances which now compose the peninsula of the Cape. That peninsula, it now appears, is a wall of granite, highest at the northern extremity, and lowering gradually to the south; faced, at its base, with grauwacky [sic], and covered, at its top, with a platform of horizontal sandstone. The penetration of the killas or grauwacky, by veins from the mass of granite which it surrounds, proves that the killas, though the superior rock, is of older formation than the granite. The granite, therefore, is a mineral that has come up from below into the situation it now occupies, and is not one of which the materials have been deposited by the sea in any shape, either mechanical or chemical. It is a species, therefore, of subterraneous lava, and the progeny of that active and powerful element, which we know, from the history both of the present and the past, has always existed in the bowels of the earth. The introduction of the granite into the situation it now occupies, must have taken place while the whole was deep under the level of the sea; this is evident from the covering of sandstone which lies on the granite, to the thickness of 1500 feet; for there can be no doubt whatever that this last was deposited by water. After this deposition, the whole must have been lifted up, as Captain Hall supposes, with such quietness and regularity, and in so great a body, as not to disturb or alter the relative position of the parts. Thus the granite is shewn, I think with great probability, to be newer than one of the rocks incumbent on it, and older than the other. I know not that we have ever before had an example of a fact which so directly ascertains the place which granite really occupies, in respect of the other parts of the mineral kingdom; it is one that from analogy might be expected to take place, and it is highly favourable to the opinion, that granite does not derive its origin from aqueous deposition. It seems, indeed, to be an instantia crucis, with respect to the two theories concerning the formation of rocks’’ (Playfair & Hall 1813, pp. 277–278). The intrusive nature of the granite on the lower slopes of Table Mountain, as depicted by Playfair & Hall (1813) (Fig. 7) was the ﬁrst published example of intrusive granite outside of Scotland, where Hutton (1795) had originally described intrusive granites. Clarke Abel (1818) had made geological observations at the Cape in 1816 and 1817, on his way to and from China. He was the ﬁrst to record the granite–schist contact along the coast near Green Point and Sea Point in Cape Town, and he illustrated the complex relationships in the contact zone in a series of very accurate drawings (Figs 8, 9, 10). He reached similar conclusions to Hall and Playfair concerning the intrusive origin of the granite and its injection into the greywackes, which he called ‘schistus’, and he cited contradictions with the Neptunian view put forward by Jameson (1808). However, he then regarded the overlying sandstones of Table Mountain as having been precipitated from the ocean in the manner advocated by the Neptunians. Abel thus concluded that ‘‘the

Figure 7 Sketch showing granite veins (white) intruding ‘killas’ or schists at Platteklip Gorge, from Playfair & Hall (1813).

mountains at the Cape of Good Hope exhibit phenomena illustrative and conﬁrmative of certain positions of both the Huttonian and Wernerian theories.’’ The anonymous reviewer of Abel’s book in the Quarterly Review suggested that he could have omitted the ‘‘the geological discussion on the appearances of the peninsula of the Cape, especially as they have been described more fully and more scientiﬁcally by Captain Hall in the Philosophical Transactions of Edinburgh’’ (Anonymous 1819). Robert Jameson (1774–1854), Regius Professor of Natural Philosophy at Edinburgh, had studied for two years under Werner in Freiberg, and was the principal proponent of the Neptunist school in Scotland, having been a founder of the Wernerian Natural History Society in 1808 (Geikie 1897). Although Jameson was elected president of this latter society on the 14th November, 1810 (Anonymous 1811), not everybody had a high opinion of him. Thomas Carlyle, writing to Robert Mitchell on 27th November, 1818, made the following remarks about him (Craig & Jones 1985, pp. 160–161): ‘‘I have heard Professor Jameson deliver two lectures. I am doubtful whether I ought to attend his class after all. He is one of those persons whose understanding is overburthened by their memory. Destitute of accurate science, without comprehension of mind, – he details a chaos of facts, which he accounts for in a manner as slovenly as he selects and arranges them.’’ Jameson countered the views on the intrusive origin of granite proposed by Playfair, Hall and Abel, in an article published in 1819 in the Edinburgh Philosophical Journal, which he had also founded. He reproduced, at length, their observations (Jameson 1819), and obtained further information, including a sketch, from one of his former pupils, Dr Adam of Calcutta (Fig. 4b). He summarised the positions of the Plutonists, as advocating that two of the formations, namely the slate and the sandstone, were of aqueous origin, while the third, granite, was of intrusive origin. He then continued: ‘‘We consider this explanation as unsatisfactory, and are inclined to view these rocks as of Neptunian and simultaneous formation; because they alternate with, and pass into each other, thus exhibiting the same general geognostical relations as occur in formations composed of sandstone and limestone, or of sandstone and gypsum. The junctions of the granite and gneiss, and of the sandstone and slate, do not present any species of veins,

8

SHARAD MASTER

Figure 8 Three views of granite dykes (light) intruding into ‘clay-slate’ (dark) at Sea Point, from Abel (1818). The engravings are by T. Fielding, after drawings by H. Raper, Esq.

PLUTONISM VERSUS NEPTUNISM AT THE SOUTHERN TIP OF AFRICA

9

Figure 9 Diagrams showing intrusive relationships between granite (light) and schist (dark) from Abel (1818), and modern views of similar features at Sea Point. Both engravings are by T. Fielding, after drawings made from sketches by H. Raper, Esq. (a) Xenoliths of schist in granite. (b) Granite dyke cutting schist. (c) Xenoliths of schist in granite, and thin granite dykelets intruding along schistose fabric. (d) Contact between granite and schist, with Lion’s Head in background. Photographed on 6th July 2007, during the 6th Hutton Symposium excursion to Sea Point.

or varieties of intermixtures, or of imbedded portions (fragments of the Huttonians), or convolutions, that do not occur at the junctions of universally admitted Neptunian rocks, such as limestone, claystone, gypsum, and sandstone. In short, the mountains and hills of the peninsula of the Cape of Good Hope, are to be considered as variously aggregated compounds of quartz, feldspar, and mica, and the whole as the result of one nearly simultaneous process of crystallisation.’’ There was no immediate rejoinder to Jameson’s views (Playfair had died in 1819). The Rev. Friedrich Hesse, who translated Latrobe’s journal into German, made cursory geological observations in Cape Town, including on the ‘schistus’ and the granite, without addressing the controversy (Latrobe & Hesse 1820). Dugald Carmichael (1821) made further observations on the geological structure of the Cape, which conﬁrmed some of the earlier results of Hall and Abel. Jameson et al. (1830) repeated the observations of Hall and Abel, and

added those of Carmichael, but reproduced, unchanged, Jameson’s earlier (1819) arguments about the Neptunian origin of the strata in the Cape Peninsula. Jameson et al. (1830) did, however, give two rival Plutonist positions, and admitted that the second of these may be regarded as ‘‘most in accordance with prevailing geological hypotheses’’: ‘‘At what period did the Cape rocks rise above the level of the sea? This question has been variously answered, according to the geological creed of those who have considered the subject. The Neptunians maintain, on plausible grounds, that all these rocks are crystallisations and deposites [sic] from the ancient waters of the globe, which have taken place in succession, – the granite being the ﬁrst formed, the slate and greywacke the next, and last of all, the principal portion of the sandstone; that, during the deposition of these diﬀerent rocks, the level of the ocean gradually sank; and that thus the mountains rose above its surface. The

10

SHARAD MASTER

again invaded the land, and covered it to a great depth; and that from this ocean was deposited the sandstone strata; that the sea again retired, and left exposed mountains, and chains of mountains of sandstone. Other Plutonists are of opinion that the slate, greywacke, and sandstone, were deposited, in uninterrupted succession, at the bottom of the sea; and that the whole mass of stratiﬁed matter was raised gradually or suddenly above the level of the ocean, forming mountains, chains of mountains, and table-lands, by that igneous agency which sent up the granite, and probably also the augit-greenstone rocks. This, of the two Plutonian views, is the most plausible, and indeed is that explanation which may be viewed as most in accordance with prevailing geological hypotheses.’’

Figure 10 (a) and (b) Drawings illustrating thin granite veins (light) intruding into schist (dark) at Sea Point, from Abel (1818). (c) Photograph showing thin granite veins (light) intruding schist (dark), at Sea Point.

Plutonians, or the supporters of the igneous origin of the granular crystallised rocks, view the formation in a diﬀerent manner. Some of the advocates of the igneous system maintain that the slate was ﬁrst deposited in horizontal strata, at the bottom of the sea, – that these strata were afterwards softened by heat, and raised from their original horizontal to their present highly inclined position, by the action of ﬂuid granite rising from the interior of the earth; and that in this way the granite and slate mountains were elevated above the sea: that the sea

It was revealed by Jameson et al. (1830) that Clarke Abel, together with his local guide, Captain Wauchope of the Royal Navy, as well as Dr Adam of Calcutta, and the recently deceased Dugald Carmichael, had all been former pupils of Dr Jameson in Edinburgh. George Champion (1836) further introduced Cape geology and topography to an American audience, following on American editions of the work by Jameson et al. (1830). In 1837, the German mineralogist Johann Friedrich Ludwig Hausmann (1782–1859) published a ‘‘contribution to the knowledge of the geognostical constitution of South Africa’’, in which he discussed the geological observations of his compatriot Hesse (Latrobe & Hesse 1820), and Hall (Playfair & Hall 1813) concerning the schists and greywackes at the foot of Table Mountain, and compared them to his observations of similar schists in the Harz mountains. Being clearly in the Neptunian camp, Hausmann (1837) attempted a correlation with the Wernerian global stratigraphy and tentatively tried to assign the schists to the ‘U } bergangsgebirge’ or ‘Transitional beds’, and the overlying sandstones of Table Mountain to the ‘Flo¨tzgebirge’ (mechanical sediments, but partly of chemical origin) (Adams 1938; Hallam 1983). In 1839, the German geologist and palaeontologist Ferdinand Krauss published an accurate account of the geology of Table Mountain and the Cape Peninsula, in which he noted, in addition to the rock types described by previous workers, the presence of several dolerite dykes which cut across both the Cape Granites and the schists and greywackes that they intruded (Krauss 1839). The Rev. William Branthwaite Clarke (1798–1878), who was a proliﬁc researcher regarded as the ‘Father of Australian Geology’ (Grainger 1982), passed through the Cape in March and April 1839, and studied its geology. In spite of the many excellent descriptions that preceded him, Clarke (1841) commenced his paper on the geological phenomena in the vicinity of Cape Town by stating that ‘‘having derived no advantage from the labours of previous geologists, his remarks must be regarded as independent of any prior description.’’ While admitting that the granite was intrusive into the schists, Clarke (1841) even suggested that the granite had intruded into the overlying sandstones. ‘‘These phænomena’’, he stated, ‘‘clearly establish the induction, that though the periods may have been distant, the schistose rocks owe their elevation to the up-burst of the granite before the deposition of the sandstone; and that subsequently the granite has been re-heated and further elevated, carrying with it the whole area described to a higher level.’’ Despite his initially dismissive attitude towards the work of his predecessors, Clarke (1841) did later acknowledge that Clarke Abel had previously identiﬁed an intrusive dyke at Kloof. Charles Darwin had visited the Cape in 1836 on the last leg of his famous voyage on the Beagle (Compton & Singer 1958). Darwin had been a former pupil of Robert Jameson at

PLUTONISM VERSUS NEPTUNISM AT THE SOUTHERN TIP OF AFRICA

Edinburgh in 1826–1827, but had found his lectures on geology and zoology ‘incredibly dull’. Darwin added that ‘‘the sole eﬀect they produced on me was the determination never as long as I lived to read a book on Geology, or in any way to study the science’’ (Darwin 1887, p. 41). It was only under the inﬂuence of Lyell (1833), whose Principles Darwin had taken with him on board the Beagle, that Darwin re-ignited his interest in geology that had been so thoroughly snuﬀed out by Jameson (see Adams 1938, p. 226). Darwin arrived at the Cape (in ‘Simon’s Bay’) on 31st May 1836, and on the 4th June he visited the famous Paarl Mountain, which he described as ‘‘a singular group of rounded granite hills’’ (Keynes 2001, p. 425). From the 8th to 15th June 1836, Darwin examined the outcrops of the Green Point granite–schist contact (which had been described by Abel (1818)), in the course of several ‘long geological rambles’ in the company of Dr Andrew Smith, with whom he remained on friendly terms for many years afterwards (Kirby 1965). Darwin (1844) referred to the previous accounts of the geology of the Cape of Good Hope by Barrow, Carmichael, Hall and Clarke. He described in meticulous detail the junction of the granite and the clay-slate, and the presence of rafts or xenoliths of clay-slate within the granite, which preserved their uniform NW–SE cleavage. He alluded to similar observations from other areas having been advanced, e.g. by Keilhau (1843), as a ‘‘great diﬃculty on the ordinary theory of granite having been injected whilst liqueﬁed’’. Darwin (1844) continued, however: ‘‘. . . but if we reﬂect on the probable state of the lower surface of a laminated mass, like clay-slate, after having been violently arched by a body of molten granite, we may conclude that it would be full of ﬁssures parallel to the planes of cleavage; and that these would be ﬁlled with granite, so that wherever the ﬁssures were close to each other, mere parting layers or wedges of the slate would depend into the granite. Should, therefore, the whole body of rock afterwards become worn down and denuded, the lower ends of these dependent masses or wedges of slate would be left quite isolated in the granite; yet they would retain their proper lines of cleavage, from having been united, whilst the granite was ﬂuid, with a continuous covering of clay-slate.’’ It was only after Charles Darwin had published his 1836 observations on the Green Point granite–schist contact (Darwin 1844), that the Plutonist position became ﬁrmly entrenched. The ageing Professor Jameson was eventually persuaded to change his own views on the origin of granite, as he admitted at a meeting of the Geological Society of Edinburgh (Adams 1938). Also in 1844 appeared a detailed account of the geology of the Cape of Good Hope by Jules Itier, who described the basal part of Table Mountain as follows (Itier 1844, p. 961; translated by SM): ‘‘The base of Table Mountain, on the side facing Cape Town, is a very distinctive porphyritic granite which is emplaced forcibly among schistose psammites, whose beds it has dislocated during penetration by injection, and whose sedimentary texture it has more or less profoundly modiﬁed.’’ Itier (1844) further observed that the contact metamorphism in the sediments adjacent to the granite, which had transformed them into ﬁne-grained garnetiferous schists, was very similar to that he had observed in schists modiﬁed by porphyritic granites at various places in the eastern Pyrenees, notably in the valleys of Carol and Railleu.

11

Itier (1848 pp. 343–349) later expanded his account of Cape geology to include the ﬁrst published colour geological map of the Cape of Good Hope. By the time Andrew Geddes Bain published his pioneering paper on the geology of South Africa (Bain 1856), the geological history of the Cape Peninsula and Table Mountain had become pretty much ﬁxed according to the following view, obtaining from the labours of Hall, Abel, Krauss, Darwin and Itier: the schist/greywacke was ﬁrst deposited, then was folded and intruded by the Cape Granite; following erosion and peneplanation, the Table Mountain sandstones were deposited unconformably on the granite and schist/greywacke; the whole ediﬁce was then uplifted, and intruded by dolerite dykes. The standard version was further entrenched by the paper on the geology of the region around the Cape of Good Hope by the French consul in Cape Town, Francois de Castelnau (1858). However, a discordant note was struck by Henry Piers (1870), who regarded the Cape granite as having intruded laterally along the contact between the schists/greywackes at the base of Table Mountain, and the subhorizontal sandstones that overlie them. In a footnote following Piers’ (1870) article, the editor of the Cape Monthly Magazine, Professor Roderick Noble, inserted the following comments, which may be taken to satisfactorily sum up the consensus view on the geology of Table Mountain, and of the origin of the Cape Granite, a view that persists, more or less unchanged, to the present day (Noble 1870): ‘‘We insert the above communication with pleasure, partly from the careful minuteness of its observation, and partly from a desire to provoke intelligent discussion on the subject. At the same time, however, we must state that our own opinion diﬀers from that of Mr. Piers, on both of the questions which he raises . . . We think that Mr. Bain’s geology of Table Mountain is substantially accurate. It is quite true that the granite is a more recent formation (or intrusion rather) than the clay-slate; but still, in the ordinary sense of the term, it is the ‘fundamental’ rock. About a hundred yards below Platte Klip, there is a spot which Capt. Basil Hall has in a sense rendered classic. Some forty years ago, when the Neptunian and Plutonic theories were still contending for victory, the celebrated navigator ‘spotted’ this particular locality where the igneous granite thrust up two great veins vertically through the superincumbent clayslate, or as it was then called, grauwacke. This seems to us to prove conclusively that the ﬂow of the granite was not lateral, but vertical. And we have further indications to the same eﬀect on the Wynberg side and out at Joostenberg, and still more conspicuously at Paarl. . . . As to the chronology of Table Mountain, we are satisﬁed that the clay-slate is the most ancient; that in course of ages it was upheaved and tilted to its present angle, and thereafter penetrated by the vertically intrusive granite; following which, the sandstone horizontal strata were deposited in a primeval Devonian sea.’’ Note that when Noble (1870) wrote the above paragraph, the controversy between the ‘Neptunian and Plutonic theories’ had long subsided, but some forty years previously, they had been ‘contending for victory’. Such a stark typically Victorian view of the debate, in terms of winners and losers, is nowadays (see Rupke 1994, and references therein) given a more nuanced treatment, which acknowledges the beneﬁcial inﬂuence of Wernerian Neptunism in the development of stratigraphy, and some of the errors and self-promoting mythologising of the Huttonian Plutonists, and their later champions such as Lyell (1833) and Geikie (1897). In the debate on the origin of the

12

SHARAD MASTER

Cape Granites, it can be seen that Wernerian Neptunists like Clarke Abel were in agreement with Plutonists like Basil Hall, concerning the intrusive nature of the granites, whilst still regarding the overlying sandstones as being the products of marine precipitation. On the other hand, we also see how Plutonists such as William Clarke and Henry Piers, whilst believing in the intrusive nature of the granite, could also have erroneously believed these granites to be intrusive into the Table Mountain sandstones. The Table Mountain Sandstone is Ordovician in age (Theron 1984), and not Devonian, as Noble (1870) had stated. Subsequent descriptions of the Green Point/Sea Point granite–schist contact have added vast amounts of additional petrographic details (von Hochstetter 1866; Cohen 1874, 1897; Prosser 1878; Shaw 1878; Mennell 1903a, b; Rogers 1905; Hatch & Costorphine 1905; Cole 1906; Schwartz 1913; du Toit 1929; Walker 1929; Haughton 1933; Walker & Mathias 1946; Theron 1984), and, more recently, geochronological information (Armstrong et al. 1998). Despite this, our understanding of the basic geological situation in these celebrated exposures near Table Mountain (arrived at after much wrangling between the disciples of Werner and Hutton) has not changed much since the detailed descriptions of Darwin and Itier in 1844. The classic geology of the Cape Peninsula has played an important part in our emerging understanding of granite intrusions, contact metamorphism and sea level changes, during the emerging years of the discipline of geology, when Neptunists and Plutonists contested at the southern tip of Africa.

3. Acknowledgements I thank the librarians Margaret Northey and Peter Duncan at the Africana Library, University of the Witwatersrand, for their considerable help in tracking down obscure references. I am grateful to Vicki Hammond for her editorial input. Several portraits were obtained from Wikimedia Commons*. The names of original artists have been given, where traceable. All modern photographs are by the author. *Editor’s note: reproduction permission for these has been sought where possible. The RSE apologies for any omissions, and will acknowledge these if further information is received.

4. References Abel, C. 1818. Narrative of a Journey into the Interior of China, and of a Voyage to and from that Country in the years 1816 and 1817. London: Longman, Hurst, Rees, Orme and Brown. Adams, F. D. 1938. The Birth and Development of the Geological Sciences. Baltimore, Maryland: Williams and Wilkins. Anderson, W. 1778. An account of a large Stone near Cape Town. In a letter from Mr. Anderson to Sir John Pringle, Bart. PRS; with a Letter from Sir William Hamilton, KB, FRS to Sir John Pringle, on having seen pieces of the said Stone. Philosophical Transactions of the Royal Society, London 68, 102–6. Anonymous 1811. Proceedings of the Wernerian Natural History Society. Scots Magazine and Edinburgh Literary Miscellany 73, 886. Anonymous 1819. Narrative of a Journey into the Interior of China, and of a Voyage to and from that Country in the years 1816 and 1817; containing an account of the most interesting transactions of Lord Amherst’s embassy to the court of Pekin, and observations on the countries which it visited. By Clarke Abel, FLS. London. 1818. Quarterly Review, New York 21, January & April 1819, 67–91. Anonymous 2006. Signiﬁcant Scots: Captain Basil Hall. http:// www.electricscotland. com/history/other/hall_basil.htm Armstrong, R., de Wit, M. J., Reid, D., York, D. & Zartman, R. 1998. Cape Town’s Table Mountain reveals rapid Pan-African uplift of its basement rocks. Journal of African Earth Sciences 27 (1A), 10.

Bain, A. G. 1856. On the Geology of South Africa. Transactions of the Geological Society, London 7 (4), 175–92. Barrow, J. 1801. An account of travels into the interior of Southern Africa in the years 1797 and 1798. London: T. Cadell Jun. and W. Davies. Brink, C. F. 1778. Nieuwste en beknopte beschryving van de Kaap de Goede Hoop; nevens een dag-verhaal van eenen landtogt, naar het binnenste van Afrika, door het Land der kleine en groote Namacquas. Amsterdam: H. J. Schneider. Buﬀon, Cte de 1774. Oeuvres comple`tes, Tome Premier. The´orie de la Terre. Paris: Panckoucke. Burnet, T. 1691. Sacred Theory of the Earth. London: R. Norton. Carmichael, D. 1819–1821. On the geological structure of part of the Cape of Good Hope. Transactions of the Geological Society, London 1 (5), 614–16. Castelnau, F. de 1858. Lettre sur la constitution ge´ologique de quelques cantons voisins du Cap de Bonne Espe´rance. Comptes Rendus de l’Acade´mie des Sciences, Paris 46, 56–7. Champion, G. 1836. Remarks on the topography, scenery, geology, &c., of the vicinity of the Cape of Good Hope. American Journal of Science 29, 230–6. Clarke, Rev. W. B. 1841. On the Geological Phenomena in the vicinity of Cape Town, South Africa. Proceedings of the Geological Society, London 3 (2), 418–23. Cohen, E. 1874. Geognostisch-petrographische Skizzen aus Su¨dafrika. Neues Jahrbuch fu¨r Mineralogie, Geologie und Palaeontologie, Stuttgart 1874, 460–505. Cohen, E. 1897. Turmalinhornfels aus der Umgebung von Capstadt. Tschermaks mineralogisches und petrographisches Mittheilungen, Wien 17, 287–8. Cole, G. A. V. 1906. On the marginal phenomena of granite domes. Geological Magazine 3, 80. Compton, A. W. & Singer, R. 1958. Darwin’s visit to the Cape. Quarterly Bulletin of the South African Library 13 (1), 9–11. Compton, J. S. 2004. The rocks and mountains of Cape Town. Cape Town: Double Storey Books. Craig, G. Y. & Jones, E. J. (eds) 1985. A Geological Miscellany. Princeton: Princeton University Press. Da Silva, L. C., Gresse, P. G., Scheepers, R., McNaughton, N. J., Hartmann, L. A. & Fletcher, I. 2000. U–Pb and Sm–Nd age constraints on the timing and sources of the Pan-African Cape Granite Suite, South Africa. Journal of African Earth Sciences 30, 795–815. Darwin, C. 1844. Geological observations on the Volcanic Islands visited during the voyage of H.M.S. Beagle, together with some brief notes on the geology of Australia and the Cape of Good Hope. London: Smith Elder. Darwin, F. (ed.) 1887. The Life and Letters of Charles Darwin, Vol. 1. London: John Murray. Degrandpre´, L. M. J. O’H. 1801a. Voyage a la Coˆte Occidentale d’Afrique, fait dans les anne´es 1786 et 1787; Suivi d’un Voyage fait au cap de Bonne-Espe´rance, contenant la description militaire de cette colonie, 2 volumes. Paris: Dentu. Degrandpre´, L. 1801b. Notes du Chapitre Premier, pp. 91–103. Notes du Chapitre Deux, pp. 194–199. In Barrow, J. (1801) Voyage dans la partie me´ridionale de l’Afrique; fait dans les anne´es 1797 et 1798. Tome Premier. Traduit de l’anglais par L. Degrandpre´. Paris: Dentu. Delisle de Sales 1779. Histoire du Monde Primitif ou des Atlantes, 3 volumes. Paris: (no name, Gay et Gide?). du Toit, A. L. 1929. The Geology of South Africa. Edinburgh: Oliver & Boyd. Forbes, V. S. 1965. Pioneer Travellers of South Africa. A geographical commentary upon notes, records, observations and opinions of travellers at the Cape 1750–1800. Cape Town: A. A. Balkema. Forbes, V. S. 1977. Some scientiﬁc matters in early writings on the Cape. In Brown, A. C. (ed.) A History of Scientific Endeavour in South Africa. Cape Town: Royal Society of South Africa. Geikie, A. 1897. The Founders of Geology. London: Macmillan. Grainger, E. 1982. The Remarkable Reverend Clarke – the life and times of the Father of Australian Geology. Melbourne: Oxford University Press. Hall, Capt. B. 1831. Fragments of Voyages and Travels, including Anecdotes of a Naval Life: Chiefly for the Use of Young Persons, Volume 1. Edinburgh: Robert Cadell. Hall, Capt. B. 1861. Lieutenant and Commander: being autobiographical sketches from his own career, from Fragments of Voyages and Travels. London: Bell & Daldy; Sampson, Low, Son & Co. Hall, J. 1798. Experiments on whinstone and lava. Transactions of the Royal Society of Edinburgh 5, 43–66.

PLUTONISM VERSUS NEPTUNISM AT THE SOUTHERN TIP OF AFRICA Hallam, A. 1983. Great Geological Controversies. Oxford: Oxford University Press. Hamilton, W. 1778. Letter from Sir William Hamilton, K. B. F. R. S. to Sir John Pringle. Philosophical Transactions of the Royal Society, London 68, Part I, 106. Hatch, F. H. & Costorphine, G. S. 1905. The Geology of South Africa. London: Macmillan and Co. Hausmann, J. F. L. 1837. Beitra¨ge zur Kunde der geognostischen Constitution von Su¨dafrika. Go¨ttinger Geologischer Anzeige, Go¨ttingen 1837, 1449–62; Neues Jahrbuch fu¨r Mineralogie, Geognosie, Geologie und Petrefaktenkunde, Stuttgart 1838, 181–7. Haughton, S. H. 1933. The Geology of Capetown and adjoining country. Explanation of Sheet No. 247 (Capetown). Pretoria: Geological Survey of South Africa. Herbert, T. 1634. Some Years Travels into Divers Parts of Africa and Asia the Great. London: R. Everingham, R. Scot, T. Beard, J. Wright & R. Chiswell. Hochstetter, F. von 1866. Beitra¨ge zur Geologie des Caplandes. Novara Expedition, geologischer Theil. II. Band, I. Abth., Wien, 19–38. Neues Jahrbuch fu¨r Mineralogie, Geologie und Palaeontologie, Stuttgart 1866, 474–5. Hutton, J. 1795. Theory of the Earth with Proofs and Illustrations. Edinburgh: Creech. Itier, J. 1844. Notice sur la constitution ge´ologique du Cap de Bonne-Espe´rance. Comptes Rendus de l’Acade´mie des Sciences, Paris 19, 960–70. Itier, Jules 1848. Journal d’un voyage en Chine en 1843, 1844, 1845, 1846, Vol. 1. Paris: Dauvin et Fontaine. Jameson, R. 1808. System of Mineralogy, comprehending oryctognosy, geognosy, mineralogical chemistry, mineralogical geography, and economical mineralogy, Vol. 3. Edinburgh: William Blackwood. Jameson, R. 1819. On the geognosy of the Cape of Good Hope. Edinburgh Philosophical Journal 1 (2), October 1819, 283–9. Jameson, R., Wilson, J. & Murray, H. 1830. Narrative of Discovery and Adventure in Africa, from the earliest ages to the present time: with illustrations of Geology, Mineralogy and Zoology. Edinburgh: Oliver & Boyd. Keilhau, M. 1843. Theory on Granite. Edinburgh New Philosophical Journal 24, 402. Kennedy, R. F. 1954. An American hero in Table Bay. Africana Notes and News 11 (3), 66–8. Keynes, R. D. (ed.) 2001. Charles Darwin’s Beagle Diary. Cambridge: Cambridge University Press. Kirby, P. R. 1965. Sir Andrew Smith, M.D., K.C.B. His Life, Letters and Works. Cape Town: A.A. Balkema. Kolb, P. 1719. Caput Bonae Spei hodiernum: das ist vollsta¨ndige Beschreibung des afrikanische Vorgebu¨rges der Guten Hofnung. Nu¨rnberg: P. C. Monath. Krauss, F. 1839. Briefwechsel. Mittheilungen an den Geheimenrath v. Leonhard gerichtet. Capstadt, 21. Juli 1838. Neues Jahrbuch fu¨r Mineralogie, Geognosie, Geologie und Petrefaktenkunde, Stuttgart 1839, 61–3. La Caille, l’Abbe´ de 1763. Journal Historique du Voyage fait au Cap de Bonne-Espe´rance. Paris: Guillyn. Latrobe, C. J. & Hesse, F. 1820. Des Evangelischen Predigers C. J. Latrobe Tagebuch einer Besuch-Reise nach Su¨d-Afrika in den Jahren 1815 und 1816. Halle und Berlin: Buchhandlung des Hallischen Waisenhauses. Le Vaillant, F. 1795. Second Voyage dans l’Interior de l’Afrique par le Cap de Bonne-Espe´rance; dans les anne´es 1783, 84 et 85, Tome I. Paris: H. J. Jansen et Cie. Lichtenstein, H. 1811–12. Reisen im Su¨dlichen Afrika in den Jahren 1803, 1804, 1805 und 1806, 2 vols. Berlin: C. Salfeld. Lyell, C. 1833. The Principles of Geology. London: J. Murray. Masson, F. 1776. Mr Masson’s Botanical Travels. An account of three journeys from the Cape into the interior parts of Africa. Philosophical Transactions of the Royal Society, London 66 (1), 268–317. Masson, F. 1994. Francis Masson’s account of Three Journeys at the Cape of Good Hope 1772–1775. With an Introduction and annotations by Frank R. Bradlow. Cape Town: Tablecloth Press. Mennell, F. P. 1903a. Minerals of some South African granites. Report of the South African Association for the Advancement of Science, First Meeting, Cape Town, 282–5. Mennell, F. P. 1903b. The minerals of some South African granites. Geological Magazine, New Series, Decade IV 10, 345–7. Mentzel, O. 1785–1787. Vollsta¨ndige und zuverla¨ssige geographische und topographische Beschreibung des beruhmten und aller Betrach-

13

tung so merkwurdigen afrikanischen Vorgebirges der Guten Hofnung, 2 volumes. Glogau: Christian Friedrich Gu¨nther. Milbert, J. 1812. Voyage pittoresque a` L’ıˆle de France, au Cap de Bonne-Esperance, 2 volumes. Paris: A. Napveu. Milbert, J. 1825. Milbert’s Reise nach Isle-de-France, dem Vorgebirge der Guten hofnung und der Insel Teneriﬀa. Edited and translated by Joh. Georg Rudolph Blumhof. Frankfurt am Main: Verlag von Franz Barrentrapp. Moore, J. 1994. Geology and climate of Table Mountain. In Hey, D. (ed.) The Mountain: An Authoritative Guide to the Table Mountain Chain, 31–42. Cape Town: Tafelberg. Noble, R. 1870. Editorial Footnote. Cape Monthly Magazine, Cape Town 2(1), 255. Piers, H. W. 1870. The Geology of Table Mountain. Cape Monthly Magazine, Cape Town 2 (1), 253–5. Playfair, J. 1802. Illustrations of the Huttonian Theory of the Earth. Edinburgh: Creech. Playfair, J. & Hall, B. 1813. Account of the structure of Table Mountain and other parts of the peninsula of the Cape; drawn up by Prof. Playfair, from the observations made by Capt. Basil Hall, RN, FRS Edin. Transactions of the Royal Society of Edinburgh 7, 269–78. Prosser, W. 1878. The Granites and Gneiss of the Colony. Transactions of the South African Philosophical Society, Cape Town 1 (9), 93–100. Raven-Hart, Maj. R. 1967. Before Van Riebeeck. Callers at South Africa from 1488 to 1652. Cape Town: C. Struik. Rogers, A. W. 1905. An Introduction to the Geology of Cape Colony. London: Longmans, Green & Co. Rupke, N. A. 1994. A second look: C. C. Gillespie’s Genesis and Geology. Isis 85 (9), 261–70. Schwartz, E. H. L. 1913. The Sea Point granite-slate contact. Transactions of the Geological Society of South Africa 16, 33–8. Shaw, J. 1878. The Petrography of Table Mountain Valley. Transactions of the South African Philosophical Society, Cape Town 1 (6), 55–65. Smith, Rev. C. 1831. Biographical notice of the late Captain Dugald Carmichael. In Hooker, W. J. (ed.) Botanical Miscellany 2, 1–59; 258–89. London: John Murray. Sonnerat, P. 1782. Voyage aux Indes orientales et a` la Chine, fait par ordre du roi depuis 1774 jusqu’en 1781, 2 volumes. Paris: Sonnerat, Froulle´, Nyon, Barrois. Sparrman, A. 1783. Resa till Goda Hopps-Udden, So¨dra Pol-kretsen och omkring Jordklotet samt till Hottentott och Caﬀer-Landen, a˚re 1772–76. Stockholm: Anders J. Nordstrom. Sparrman, A. 1785. A Voyage to the Cape of Good Hope, towards the Antarctic Polar Circle, and round the world; but chiefly into the country of the Hottentots and Caﬀres, from the year 1772 to 1776, 2 volumes. Dublin: White, Cash & Byrne. Tachard, G. 1686. Voyage de Siam des Pe`res Je´suites, enyoyez par le Roy aux Indes & a` la Chine, avec leurs Observations Astronomiques, et leurs Remarques de Phisique, de Ge´ographie, d’Hydrographie, & d’Histoire. Paris: Seneuze et Horthemels. Theron, N. J. 1984. The Geology of Cape Town and environs. Explanation of Sheets 3318 CD and DC, and 3418 AB, AD and BA. Pretoria: Geological Survey of South Africa. Thunberg, C. P. 1788. Resa uti Europa, Africa, Asia fo¨rra¨ttad a˚ren 1770–1779, vol. 1, 1788; vol. 2, 1789; vol. 3, 1791; vol. 4, 1793. Uppsala: Joh. Edman (vols. 1–3) & Joh. Edmans Enka (vol. 4). Thunberg, C. P. 1986. Carl Peter Thunberg: Travels at the Cape of Good Hope 1772–1775. Edited by V.S. Forbes. Cape Town: Van Riebeeck Society. Valentyn, F. 1726. Beschryving van ’t Nederlandsch Comptoir op de Kust van Malabar, en van onzen Handel in Japan, mitsgaders een Beschryving van Kaap der Goede Hoope en ’t Eyland Mauritius, met de zaaken tot de voornoemde Ryken en Landen behoorden. Dordrecht: Johannes van Braam. Walker, A. R. E. 1929. The Sea Point granite-slate contact. 15th International Geological Congress, South Africa, Guidebook A3. Walker, F. & Mathias, M. 1946. The petrology of two granite-slate contacts at Cape Town, South Africa. Quarterly Journal of the Geological Society, London 102 (4), 499–518. Werner, A. G. 1787. Kurze Klassifikation und Beschreibung der verschiedenen Gebirgsarten. Dresden: Waltherischen Hofbuchhandlung. Werner, A. G. 1791. Neue Theorie von der Entstehung der Ga¨nge, mit Anwendung auf den Bergbau besonders den freibergischen. Freiberg: Gerlach.

MS received 4 December 2008. Accepted for publication 5 December 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 15–33, 2010 (for 2009)

The sanukitoid series: magmatism at the Archaean–Proterozoic transition Herve´ Martin1a,b,c, Jean-Franc¸ois Moyen2 and Robert Rapp3 1a

Clermont Universite´, Universite´ Blaise Pascal, Laboratoire Magmas et Volcans, BP 10448, F-63000 ClermontFerrand, France

1b

CNRS, UMR 6524, LMV, F-63038 Clermont-Ferrand, France

1c

IRD, R 163, LMV, F-63038 Clermont-Ferrand, France Email: [email protected]

2

Department of Geology, University of Stellenbosch, Private Bag X 01, 7602 Matieland, South Africa

3

Research School of Earth Sciences, The Australian National University, Canberra, ACT, 0200 Australia

ABSTRACT: A speciﬁc type of granitoid, referred to as sanukitoid (Shirey & Hanson 1984), was emplaced mainly across the Archaean–Proterozoic transition. The major and trace element composition of sanukitoids is intermediate between typical Archaean TTG and modern arc granitoids. However, among sanukitoids, two groups can be distinguished on the basis of the Ti content of the less diﬀerentiated rocks of the suite: high- and low-Ti sanukitoids. Melting experiments and petrogenetic modelling show that they may have formed by either (1) melting of mantle peridotite previously metasomatised by felsic melts of TTG composition, or (2) by reaction between TTG melts and mantle peridotite (assimilation). Rocks of the sanukitoid suite were emplaced at the Archaean–Proterozoic boundary, possibly marking the time when TTG-dominated granitoid magmatism changed to a more modern-style, arc-dominated magmatism. Consequently, the intermediate character of sanukitoids is not only compositional but chronological. The succession of granitoid magmatism with time is integrated in a plate tectonic model where it is linked to the thermal evolution of subduction zones, reﬂecting the progressive cooling of Earth: (1) the Archaean Earth’s heat production was high enough to allow the production of large amounts of TTG granitoids formed by partial melting of recycled basaltic crust (‘slab melting’); (2) at the end of the Archaean, due to the progressive cooling of the Earth, the extent of slab melting was reduced, resulting in lower melt:rock ratios. In such conditions the slab melts can be strongly contaminated by assimilation of mantle peridotite, thus giving rise to low-Ti sanukitoids. It is also possible that the slab melts were totally consumed in reactions with mantle peridotite, subsequent melting of this ‘melt-metasomatised mantle’ producing the high-Ti sanukitoid magmas; (3) after 2·5 Ga, Earth heat production was too low to allow slab melting, except in relatively rare geodynamic circumstances, and most modern arc magmas are produced by melting of the mantle wedge peridotite metasomatised by ﬂuids from dehydration of the subducted slab. Of course, such changes did not take place exactly at the same time all over the world. The Archaean mechanisms coexisted with new processes over a relatively long time period, even if they were subordinate to the more modern processes. KEY WORDS: geochemistry, granitoid, magma/melt interactions, petrogenesis, slab melting, temporal change in magma production The genesis of the continental crust started very early in Earth’s history: indeed, detrital zircons from Jack Hills, in Western Australia record the existence of 4·40 Ga granitic (s.l.) crust (Wilde et al. 2001). Whilst the ﬁrst half of Earth history mainly corresponds to the extraction of juvenile crust from the mantle, recycling mechanisms existed before 4·0 Ga ago (Cavosie et al. 2004, 2005, 2006; Watson & Harrison 2005; Harrison & Schmitt 2007; Blichert-Toft & Albare`de 2008), but were highly subordinated processes. Due to the greater Earth heat production (Brown 1985), the petrogenetic processes that operated were diﬀerent from modern ones, resulting in the genesis of unique lithologies such as komatiites and massive volumes of tonalite trondhjemite and granodiorite (TTG) magmas (Viljoen & Viljoen 1969; Glikson 1971; Windley & Bridgwater 1971; Arth & Hanson 1972; Barker & Arth 1976;

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016120

McGregor 1979; Condie 1981; Jahn et al. 1981; Martin et al. 1983). Based on petrological and experimental studies, as well as on geochemical modelling, the genesis of Archaean TTG has been explained by partial melting of hydrous basalt metamorphosed into garnet-bearing amphibolite or eclogite (Barker & Arth 1976; Martin 1986, 1987, 1993, 1994; Rapp et al. 1991, 2003; Rapp & Watson 1995; Martin et al. 1997, 2005; Foley et al. 2002; Martin & Moyen 2002). In contrast, it is more seldom proposed that TTG are generated by the extensive fractional crystallisation of water-rich basalt in a subduction environment (Kamber et al. 2002; Kleinhanns et al. 2003). If most researchers consider that TTG were generated by the melting of hydrated basalt, they disagree about the exact site where this melting took place. The two end-member possibilities are: (1) partial melting of basalt which underplated

16

HERVE MARTIN ET AL.

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION

a thickened crust (Atherton & Petford 1993; Rudnick 1995; Albare`de 1998; de Wit 1998; Smithies 2000; Smithies et al. 2005; Be´dard 2006); (2) a subducted hot oceanic slab that melted rather than dehydrating (Martin 1986; Condie 1989; Rollinson 1997; Barth et al. 2002; Foley et al. 2002; Kamber et al. 2002; Martin & Moyen 2002; Rapp et al. 2003; Condie 2005; Nair & Chacko 2005; Martin et al. 2008). After the end of the Archaean (2·5 Ga), and until today, most of the juvenile continental crust is formed by melting of a ﬂuid metasomatised peridotite followed by diﬀerent degrees of diﬀerentiation, generating the BADR (Basalt Andesite Dacite Rhyolite) suites typical of subduction environments. There, the source of the magmas is considered as being the mantle wedge peridotite metasomatised by ﬂuids resulting from the dehydration of the subducted slab (Tatsumi 1989; Pawley & Holloway 1993; Liu et al. 1996; Schmidt & Poli 1998; Forneris & Holloway 2003). The transition between TTGs and BADRs roughly took place at the Archaean–Proterozoic transition, about 2·5 Ga ago. At the same period, high-Mg dioritic, tonalitic and granodioritic magmatic rocks were generated and emplaced into all Archaean cratons. These plutons, commonly called late granodioritic or granitic plutons, were ﬁrst identiﬁed by Shirey & Hanson (1984), who referred to them as Archaean sanukitoids. These rocks are now found in most Late Archaean terranes (2·9–2·5 Ga) (Shirey & Hanson 1984; Stern 1989; Stern & Hanson 1991; Smithies & Champion 1999; Moyen et al. 2001b, 2003); they possess both modern (classical calcalkaline diﬀerentiation, similar to BADR association, high transition element contents) and Archaean (low HREE contents, strongly fractionated REE patterns, etc. . . .) geochemical characteristics. The transitional character of sanukitoids is not only compositional but also chronologic, being emplaced during the ‘hinge’ period between two epochs dominated by TTG (Archaean) and BADR (Proterozoic/Phanerozoic) juvenile crustal magmatism. Consequently, their study could provide not only new insights into the change in petrogenetic mechanisms during this period, but also into the changing geodynamics on Earth across the w2·5 Ga boundary. The purpose of this paper is: (1) to review the geochemical and petrologic characteristics of sanukitoids; (2) to address their petrogenesis; (3) to discuss possible geodynamic environments for their generation; and (4) to consider their temporal distribution over the whole of Earth’s crustal evolution.

1. Sanukitoids 1.1. Definition Shirey & Hanson (1984) ﬁrst recognised a suite of Late Archaean felsic intrusive and volcanic rocks in the Superior Province that had both mineralogical and chemical characteristics clearly diﬀerent from TTG, which had up until then been viewed as, volumetrically, the overwhelmingly dominant granitoid throughout the Archaean. Because the major element geochemistry of these rocks resembled that of Miocene high-Mg Andesite (Sanukite) from the Setouchi volcanic belt

17

of Japan (e.g. Tatsumi & Ishizaka 1982), Shirey & Hanson (1984) referred to them as ‘Archaean sanukitoids’. Since this pioneering work, sanukitoids have been described in most Archaean terranes: the Superior Province (Shirey & Hanson 1984, 1986; Stern & Hanson 1991; Be´dard 1996; Stevenson et al. 1999), Wyoming (Frost et al. 1998), the Baltic shield (Querre´ 1985; Lobach-Zhuchenko et al. 2000, 2005, 2008; Halla 2005; Kovalenko et al. 2005; Samsonov et al. 2005; Ka¨pyaho 2006), South India (Balakrishnan & Rajamani 1987; Jayananda et al. 1995; Krogstad et al. 1995; Moyen et al. 2001b, 2003; Sarvothaman 2001), China (Jahn et al. 1998), Limpopo (Barton et al. 1992; Millonig et al. 2008), the Central Pilbara craton (Smithies & Champion 1999) and the Amazonian craton (Medeiros & Dall’Agnol 1988; Althoﬀ 1996; Leite et al. 2004). Compared with TTG, sanukitoids still represent a volumetrically subordinate component of the Archaean crust; however, they are a common component of most Late Archaean cratons.

1.2. Composition Based on ﬁeld observations, sanukitoids deﬁne a complete magmatic series, from diorites to granites (the ‘sanukitoid suite’ of Stern & Hanson 1991). The two most common rock types are: (1) medium-grained, equigranular monzodiorites to granodiorites, containing small (5–10 mm) clusters of biotite, hornblende and rare relicts of hornblende-rimmed clinopyroxene, which give the rock a very distinctive, black and white ‘spotted’ aspect; (Fig. 1a–b); (2) porphyritic monzogranite (Fig. 1d–e), with large (2–5 cm) to very large (w10 cm) phenocrysts of K-feldspar in a coarse-grained matrix. In both case, the paragenesis consists of quartz, plagioclase (An20–30), perthitic microcline, hornblende and biotite. Accessory phases are magnetite, ilmenite, epidote, sphene, apatite, zircon and allanite. Microgranular, maﬁc dioritic to monzodioritic enclaves are common (Fig. 1c, f–h); they are ﬁne grained (0·1–1 mm), with occasional K-feldspar phenocrysts with rapakivi texture. They also typically contain small maﬁc clusters of biotite with ‘spots’ of dull black amphibole. In some places, relict diopside has been observed within amphibole grains. Sanukitoid can occur as plutons of all sizes, with a broad range of crustal emplacement levels and degrees of heterogeneity. For example, sanukitoids in the Central Pilbara Craton (Smithies & Champion 1999) form small (<1 km), homogeneous stocks of shallowly emplaced magmas, whereas sanukitoids of South India (the ‘Closepet granite’, Jayananda et al. 1995; Moyen et al. 1997, 2001a), form a huge intrusive body that is about 400 km long and 20 km wide and is rooted in the lower granulitic crust. The porphyritic monzogranite of the Closepet batholith is associated with migmatites and anatectic granites derived from melting of the surrounding TTG basement. Sanukitoid compositions range between maﬁc (SiO2w50% and MgOw8%) and felsic (SiO2w75% and MgOw0·1%) end-members (Fig. 2). This could result either from diﬀerentiation of a primitive parental magma (e.g. Reddy 1991; Stern & Hanson 1991) or from contamination by older crustal

Figure 1 Field appearance of sanukitoids: (a)–(c) medium-grained, equigranular monzodiorite (Low-Ti sanukitoid); (d)–(h), porphyritic monzogranite (High-Ti sanukitoid). (a) Typical sanukitoid amphibole granodiorite; 2·55 Ga ‘Dod gneisses’, Kolar belt, India (Krogstad et al. 1991; Chardon et al. 2002). (b) Biotiteamphibole bearing granodiorite in the 2·67 Ga Matok Pluton, Limpopo Belt, South Africa (Kreissig et al. 2001). (c) Maﬁc microgranular enclave-bearing sanukitoid of the Rio Maria terrane, Amazonian Craton, Brazil (courtesy R. Dall’Agnol) (Medeiros & Dall’Agnol, 1988). (d) Porphyritic monzogranite from the Closepet batholith, South India. (e) Porphyritic monzogranite in the 2·57 Ga Bulai Pluton, Limpopo Belt, South Africa (Barton et al. 1992). (f). Stretched microgranular maﬁc enclave in the Closepet batholith. (g) Pillowed (monzo-) dioritic enclave, Closepet. (h) Mingling relationships between a porphyritic monzogranite and monzodiorites, Matok.

18

HERVE MARTIN ET AL.

Figure 2 Harker diagrams for both low-Ti (ﬁlled circles) and high-Ti (open squares) sanukitoids.

components during emplacement (Moyen et al. 2001a). For the Closepet batholith for example, it has been demonstrated (Jayananda et al. 1995; Moyen et al. 1997, 2001a; among others) that the linear trends for both major and trace elements, over a wide variation of SiO2 contents (from w50% up to 75%) are the result of crustal contamination. In this case, mantle-derived magmas interacted with the crust in which they intruded. Their mantle characters were obliterated and altered by a superimposed crustal signature, such that they were often considered to have a mixed origin.

The details of this ‘late’ evolution are not the focus of this paper, which deals with the origin of the juvenile component, regardless of its subsequent evolution. Consequently, this paper will only refer to the less diﬀerentiated component of the sanukitoid suites, thus excluding all samples with SiO2 >62%, that are assumed to be strongly modiﬁed by either interaction with a felsic crustal component, or by diﬀerentiation of these primary magmas. The diﬀerences between the two main petrologic types identiﬁed above (medium-grained, equigranular monzodiorites

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION

19

Figure 3 TiO2 vs. MgO plot showing that for the same degree of diﬀerentiation (anti-correlated with MgO), TiO2 content is signiﬁcantly higher in porphyritic monzogranite (open squares) than in medium-grained, equigranular monzodiorites (ﬁlled circles), thus supporting the discrimination between two groups of sanukitoids (Low-Ti and high Ti).

Figure 5 Chondrite normalised REE patterns for average Archaean TTG (grey triangles), low-Ti (ﬁlled circles) and high-Ti (open squares) sanukitoids with SiO2 <62%. The grey ﬁeld represents the compositional ﬁeld for sanukitoids. Normalisation values are from Masuda et al. (1973) divided by 1·2.

Figure 4 Cationic Ca–Na–K diagram showing that both the low-Ti (ﬁlled circles) and high-Ti (open squares) sanukitoids are diﬀerent of Archaean TTG (grey ﬁeld, Martin et al. 2005). They show a K-enrichment, thus occupying a position intermediate between TTG and the classical calc-alkaline trend (CA) as deﬁned by Barker & Arth (1976). It must also be noted that the high-Ti sanukitoids are slightly K-richer than the low-Ti ones.

and porphyritic monzogranite), are not only textural, but also correspond to geochemical diﬀerences. Figure 2 shows that for some elements (e.g., Al2O3, Na2O and K2O), chemical trends are exactly the same in the two groups, but for other elements the geochemical evolution diﬀers, particularly for the less diﬀerentiated rocks. This is well exempliﬁed by the TiO2 vs. SiO2 (Fig. 2) or TiO2 vs. MgO (Fig. 3) plots, where for the same degree of diﬀerentiation, TiO2 content is signiﬁcantly higher in porphyritic monzogranite than in medium-grained, equigranular monzodiorites. Consequently, it is proposed to discriminate these two sanukitoid types on the basis of their TiO2 content: in this paper, the medium-grained, equigranular monzodiorites will subsequently be referred to as low-Ti sanukitoids, whereas the porphyritic monzogranites will be referred as high-Ti sanukitoids. In a K–Na–Ca cationic triangle (Fig. 4), sanukitoids do not show any aﬃnity with Archaean TTG (Martin et al. 2005) and rather show an evolution trend similar to the classical calcalkaline diﬀerentiation trend of Barker & Arth (1976). It must

Figure 6 Primitive mantle (Sun & McDonough 1989) normalised multi element diagram for average Archaean TTG (grey triangles), both low-Ti (ﬁlled circles) and high-Ti (open squares) sanukitoids with SiO2 <62%. The grey ﬁeld represents the compositional ﬁeld for sanukitoids.

be noted that the low-Ti sanukitoids are slightly poorer in K and richer in Na than their high-Ti equivalents. Table 1 reports the average composition of 104 sanukitoids from the literature. Out of the published data, only samples with SiO2 lower than 62% were kept, as they are considered to be representative of the maﬁc pole of the diﬀerentiation suites – regardless of the diﬀerentiation mechanism involved. These 104 analyses were classiﬁed into 57 low-Ti- and 47 high-Ti sanukitoids, whose average composition is also given. The sanukitoids are meta-aluminous (A/CNK=0·79) and moderately potassic (K2O/Na2O=0·72). Mg# (molecular Fe/ (Fe+Mg)) is quite high (0·53), as are Ni and Cr contents (54 and 104 ppm, respectively). Sr and Ba are typically greater than 1000 ppm (1108 and 1471 ppm respectively); Na2O (4·31%) and K2O (3·11%), (as well as most LILE) contents are also high. Similarly, LREE (e.g. LaN =234) contents are high

20

HERVE MARTIN ET AL. Table 1 Average composition and standard deviation for 104 sanukitoids from the literature. Only samples with SiO2 lower than 62% were taken into consideration, as they are considered as representative of the maﬁc pole of the diﬀerentiation suites, regardless of the diﬀerentiation mechanism that operated. These 104 analyses fall into two groups: 57 are low-Ti- and 47 are high-Ti sanukitoids. The average compositions of 250 modern arc granitoids (Martin 1994) and 1094 TTGs (Martin et al. 2005) are also given for comparison. Fe2O3*=Total iron expressed as Fe2O3; Mg#=molecular ratio Mg/(Mg+Fe); A/CNK=molecular ratio Al/(Ca+Na+K). TTG (n=1094)

Sanukitoid <62% SiO2 (n=104)

Low-Ti High-Ti Modern Arc Granitoids Sanukitoid Sanukitoid (n=250) <62% SiO2 (n=57) <62% SiO2 (n=47)

Average Std. dev. Average Std. dev. Average Std. dev. Average Std. dev. wt.% SiO2 Al2O3 Fe2O3* MnO MgO CaO Na2O K 2O TiO2 P2O5 ppm Rb Ba Nb Sr Zr Y Th Ni Cr V La Ce Nd Sm Eu Gd Dy Er Yb Lu K2O/Na2O Mg# A/CNK (La/Yb)N

Average

69·51 15·59 3·24 0·05 1·25 3·16 4·72 1·95 0·38 0·15

3·64 1·14 1·56 0·05 0·77 1·11 0·77 0·77 0·21 0·10

58·65 16·14 6·75 0·12 3·90 5·53 4·31 3·11 0·93 0·56

3·56 1·10 1·94 0·07 1·41 1·24 0·73 0·82 0·42 0·27

59·18 16·08 6·14 0·10 4·33 5·52 4·49 3·04 0·69 0·42

3·31 1·07 1·49 0·03 1·27 1·33 0·75 0·85 0·18 0·12

58·00 16·22 7·50 0·13 3·39 5·55 4·09 3·20 1·21 0·72

3·47 1·39 1·71 0·07 1·83 1·39 0·82 0·88 0·46 0·31

68·1 15·07 4·36 0·09 1·55 3·06 3·68 3·4 0·54 0·15

66 713 7 490 135 12 7 18 40 48 31·4 57·8 22·4 4 0·9 2·4 1·7 0·76 0·64 0·13 0·41 0·43 1·00 32·4

43 465 5 217 108 16 0·6 17 75 29 23·8 257·0 17·0 2·3 0·4 1·4 0·9 0·49 0·40 0·50

87 1471 13 1108 237 27 9 54 104 118 73·9 152·4 69·7 11·7 2·8 7·8 4·4 1·97 1·60 0·26 0·72 0·53 0·79 30·5

43 601 7 512 112 14 7 33 73 43 42·2 66·7 28·1 4·7 1·1 2·9 1·8 0·88 0·78 0·12

83 1493 10 1202 172 20 9 69 143 110 57·0 120·5 56·5 9·5 2·4 6·7 3·5 1·49 1·29 0·21 0·68 0·58 0·78 29·2

37 575 3 563 55 7 5 28 60 34 17·4 34·0 16·0 2·8 0·7 2·0 1·2 0·55 0·50 0·07

92 1445 17 994 316 35 10 36 58 127 94·4 191·1 85·8 14·3 3·3 9·2 5·5 2·56 1·98 0·32 0·78 0·47 0·80 31·5

38 699 7 366 104 13 9 41 55 42 51·6 90·7 37·4 5·9 1·2 2·4 1·2 0·68 0·76 0·11

110 715 12·1 316 171 26 11·8 10·5 23 76 31 67 27 5·3 1 5·5 5·2 3 3·2 0·5 0·92 0·41 0·98 6·4

and HREE (e.g. YbN =7·7) moderately low, resulting in strongly fractionated patterns ((La/Yb)N =30·5); which do not show any signiﬁcant Eu anomaly (Fig. 5). In addition, high-Ti sanukitoids are enriched in REE relative to their low-Ti counterparts. On a primitive mantle-normalised multi-element diagram (Fig. 6), sanukitoids (both types) do not show signiﬁcant negative anomalies in Zr or Y and only a small negative one for Ti. On Harker plots, (Fig. 2), both low and high-Ti sanukitoids plot on a single trend for most elements (Al2O3, CaO, Na2O, K2O and partly FeOt), pointing to the fact that the two groups formed from a similar source by similar mechanisms. Only MgO and TiO2 depict slightly diﬀerent trends that converge towards a common silica-rich pole. This conﬁrms that the diﬀerences between the two groups do not reﬂect diﬀerentiation mechanisms, but rather correspond to distinct maﬁc

Std. dev

6·2 1·6 2 0·1 1 0·64 0·49 1·1 0·32 0·08

50 205 5 150 53 5 6·5 8 15 45 9 17 7 14 0·5 1 0·1 1 0·5 0·1

end-members. Figures 5 and 6 also show that both REE and multi-element patterns are parallel, except for a slight negative Sr anomaly in high-Ti sanukitoids that does not exist in the low-Ti group. This parallelism strongly militates in favour of a similar source for the two facies. In addition, the trace element patterns are also essentially parallel to those of average Archaean TTG, leading us to conclude that TTG are genetically related to sanukitoids in some way.

1.3. Petrogenesis The average Mg# in sanukitoids is 0·53, but can reach values >0·65 in the more maﬁc samples. These very high Mg#s and correspondingly high Cr and Ni concentrations (>100 ppm and >300 ppm respectively) in the most primitive members of the sanukitoid suites preclude a crustal (including basaltic) source; for reference, the Mg# of experimental TTG liquids

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION

21

Figure 7 (A) (Nb/Y) vs. (La/Yb) plot for low-Ti (ﬁlled circles) and high-Ti (open squares) sanukitoids with SiO2 <62%. The dark grey ﬁeld is that of arc magmas generated by melting of a peridotite metasomatised by ﬂuids. Data are from Fiji (Rogers & Setterﬁeld, 1994) and the New Hebrides (Monzier et al. 1997). The light grey ﬁeld is that of modern high-Mg andesites considered to be generated by the melting of a mantle peridotite enriched by felsic melts. Data are from Baja California (Calmus et al. 2003) and Ecuador (Bourdon et al. 2003; Hoﬀer 2008). See text for more detailed explanation. Sanukitoids point to a mantle source metasomatised by felsic melts. (B) (La/Yb)N vs. K2O plot that compares the composition of sanukitoids with that of typical arc BADR suites (Martin 1994). Sanukitoids, draw a trend parallel to that of BADR ﬁeld, thus pointing to similar petrogenetic mechanisms and source. However, the fact that the ﬁeld of Archaean TTG perfectly plots on one extremity of the sanukitoid trend indicates the possible role of a metasomatic agent played by TTG magma.

generated by partial melting of basalts never exceeds 0·45 (Rapp & Watson 1995; Zamora 2000). Similarly the SiO2 content of the less diﬀerentiated TTG is of about 60% (Martin 1994), whereas it is of about 50% in sanukitoids. Therefore, the source of sanukitoids must be ultramaﬁc. On the other hand, the same rocks are very rich in LILE; which because of the high Mg#, Cr and Ni, precludes any interpretation in terms of enrichment through fractional crystallisation, as this process would eﬃciently deplete compatible elements from the magmas (Martin & Sigmarsson 2007). Contamination of an LILE-rich felsic continental crust by komatiitic or basaltic magmas could also generate sanukitoid magmas. However, Stern (1989), and more recently Smithies & Champion (1999) modelled the interaction between maﬁc (or ultramaﬁc) melts and the felsic crust, and demonstrated that this process cannot reproduce both characteristics of primitive sanukitoids (high Mg#, Ni and Cr together with high SiO2 and LILE contents). Therefore, the present authors interpret the LILE enrichment as a primary characteristic, reﬂecting the nature of the source; that source must at least in part be ultramaﬁc (because of the high Mg#s, Cr and Ni), and LILE enriched; the only reliable and realistic possibility is that sanukitoids derive from a source formed by the interaction of a felsic (TTG) melt and ultramaﬁc rock (mantle peridotite). The sanukitoid melt itself may come about from a single stage process of melt inﬁltration, hybridisation, and assimilation (e.g., Rapp et al. 1999), or in a two-stage process in which melt metasomatism is followed by partial melting of the metasomatised peridotitic source. In a modern subduction zone, where a metasomatised and geochemically enriched mantle wedge is the main source of arc magmatism, two metasomatic agents, (both derived from crustal components of the subducting oceanic lithosphere) are generally invoked: hydrous, solute-rich, possibly supercritical ﬂuids, and hydrous, SiO2-rich (i.e., felsic) melts. These two metasomatic agents have very contrasted behaviour with respect to some chemical elements like HFSE; for instance solid/fluid solid/melt KdNb >10, whereas KdNb <1. In other words, hydrous ﬂuids are unable to transfer Nb from the slab into the mantle wedge, whereas a felsic melt can. In addition, felsic melts

generated within the garnet stability ﬁeld will have an ‘adakitic’ signature characterised by very low Y and Yb contents (Martin et al. 2005; Macpherson et al. 2006; Moyen 2009). Consequently, the nature of the metasomatic agent can be discussed in the (Nb/Y) vs. (La/Yb) diagram (Fig. 7A), where the composition of sanukitoids is compared with that of modern arc magmas. The dark grey ﬁeld is that of modern magmas from New Hebrides (Monzier et al. 1997) and Fiji (Rogers & Setterﬁeld 1994) where the magmatic signature is free of any continental contamination and where the metasomatic agent is considered to be exclusively ﬂuids produced by sub-solidus dehydration of the subducted slab. The light grey ﬁeld is that of modern high-Mg andesites and lowsilica adakites (Bourdon et al. 2003; Hoﬀer 2008), which are also considered to be the result of partial melting of melt-metasomatised mantle peridotite. Since Late Archaean sanukitoids have higher Nb/Y and La/Yb than typical modern arc magmas, a ﬂuid-metasomatised mantle source can be ruled out. On the other hand, sanukitoids overlap the ﬁeld for high-Mg andesites or low-silica adakites, magmas that are considered as generated either through partial of a mantle peridotite metasomatised by adakitic melts (Martin et al. 2005, for review), or by direct reaction (assimilation) of mantle peridotite with TTG melts (e.g., Rapp et al. 1999). In addition, Figures 5 and 6 shows that except for transition elements, the REE and multi-element patterns of sanukitoids are strictly parallel to TTG patterns, thus pointing to a strong genetic link between them. This conclusion is reinforced by the (La/Yb)N vs. K2O plot (Fig. 7B), comparing the composition of sanukitoids with typical arc BADR suites and Archaean TTGs (Martin 1994). Sanukitoids follow a trend parallel to the BADR ﬁeld, with the same range of K2O contents, (La/Yb)N being closely correlated with K2O. This reﬂects the similarity of the petrogenetic processes: possibly melting of a metasomatised mantle. However, the BADR and sanukitoid trends are parallel but not superimposed, the later having higher (La/Yb)N, indicating diﬀerent metasomatic agents; the overlap between the TTG ﬁeld and the end of the sanukitoid trend points to TTGs being a possible metasomatic agent.

22

HERVE MARTIN ET AL.

Collectively, these evidences show that the most likely metasomatic agent responsible for the peridotitic source enrichment is a TTG-like felsic melt.

2. Experimental petrology The only way to obtain appreciable volumes of LILE-enriched magmas from a peridotitic source is by enrichment of the mantle source prior to melting. This would imply a two-stage petrogenetic process for sanukitoid magmas in which melting is preceded by metasomatism, i.e., the reaction between an LILE-enriched ﬂuid or melt and mantle peridotite. As discussed above (Fig. 7), this metasomatic agent is considered as being an SiO2-rich melt. In a modern subduction zone setting, these felsic melts are presumed to be produced by partial melting of subducted basaltic oceanic crust, referred to as ‘adakites’, a term more or less synonymous with the ‘slab melt’ (henceforth referred to as ‘pristine’ adakites, to denote a slab melt that has yet to experience modiﬁcation by reaction with mantle peridotite). As Archaean TTG, the chemistry of adakites shows that they are derived by partial melting of garnetbearing maﬁc (metabasaltic) crust, followed by variable degrees of subsequent ‘interaction’ with mantle peridotite, presumably in the overlying mantle wedge (see Martin et al. 2005, for review). Given the geochemical evidence presented above for a genetic link between TTG granitoids and sanukitoids, a number of experimental studies looking at various aspects of chemical interaction between TTG liquids and peridotite have been undertaken in recent years (e.g. Sen & Dunn 1994; Yaxley & Green 1998; Rapp et al. 1999, 2006; Prouteau et al. 2001; Hoﬀer 2008). These studies focused on the equilibrium between the ‘hybridised slab melt’ (Rapp et al. 1999) and the modiﬁed peridotite assemblage. They implicitly assume a ‘one-stage’ process, in which the adakitic (or TTG) melt is transformed into a sanukitoid on its way up to the surface; sanukitoid forms contemporaneously with the geological event that generated the felsic melt (Fig. 8). Earlier experimental studies in the 1970s and 1980s examined processes of ‘melt hybridisation in subduction zones’ (Nicholls & Ringwood 1973; Sekine & Wyllie 1982; Johnson & Wyllie 1989), where ‘assimilation of mantle peridotite by an ascending slab-derived melt’ led to the formation of ‘zones of hybrid olivine pyroxenite which, in turn, could be remelted to generate a spectrum of island arc magmas’ (Nicholls & Ringwood 1973; Johnson & Wyllie 1989). This scenario immediately calls to mind a ‘two-stage’ process, in which the metasomatic event is separated in time from the melting (or remelting) event, which yields sanukitoids (Fig. 8). Deciding between a ‘one-stage’ and a ‘two-stage’ process is diﬃcult, if not impossible, on experimental grounds alone. Indeed, the only conclusions that can be drawn from experiments is that a sanukitoid melt can be formed in equilibrium with a maﬁc to ultramaﬁc mineral assemblage, which provides no constrains on the path that was followed in the P–T–X space to reach these conditions, let alone the duration of this path. A critical factor in the reaction between slab-derived, pristine melts and peridotite in the mantle wedge is the so-called ‘eﬀective melt-to-rock ratio’ (Rapp et al. 1999), which refers to the relative proportions of adakite melt and peridotite ‘rock’ involved in the ‘melt-rock reaction’ (Kelemen et al. 1993, 1998). This parameter is obviously easily controlled in laboratory experiments, but in nature, it depends upon the scale at which the interaction between slab-derived melts and the mantle wedge are considered, as well as the physical mechanism for melt inﬁltration (e.g., porous ﬂow? fracture

propagation?). The melt:rock ratio can be high; this does not mean that the mantle as a whole was soaked with melts, but rather that the scale of the interactions was such that only limited portions of the mantle were allowed to react with the melt. In other words, low melt:rock ratios correspond to interactions involving large amounts of the mantle – maybe corresponding to narrow magma pathways, or magma percolating through the mantle on a grain-scale, with a correspondingly high interface surface between melt and mantle. In contrast, high melt:rock ratios can be obtained where large magma conduits restrict the contact surface between the melt and the peridotite, and/or high magma ﬂow in large conduits reduces the time range available for interaction, and/or armouring of the edges of the conduit makes further reactions diﬃcult or impossible. Consequently, in natural geological systems, the melt:rock ratio can vary between zero and inﬁnity, as a function of the magma ﬂow rate, overall supply, and of the size and geometry of the magma pathways, and the mechanism involved in melt inﬁltration and transport. Regardless of these issues, melt–rock reactions in the overlying mantle wedge above a subducting slab can generally be assumed to be dominated by processes of assimilation when the eﬀective melt:rock ratio is high (one-stage process; Fig. 8), and by cryptic and modal metasomatic processes when the eﬀective melt:rock ratio is low. In the former case, mantlehybridised adakitic melts will be produced, and in the latter case a melt-metasomatised peridotite (sensu lato) mantle wedge is formed, which itself can then become the source for magmas by subsequent partial melting events (i.e., the ‘two-stage’ process; Fig. 8). The composition of ‘mantle-hybridised slab melts’ and the metasomatic mineral assemblages that form as a result of such melt–rock reactions in the mantle wedge, will clearly depend upon such factors as the ‘eﬀective melt:rock ratio’, pressure, temperature, and water content (all strictly controlled in laboratory experiments), and comprehensive and systematic experimental studies are lacking. Nevertheless, certain general observations can be made regarding the processes of peridotite assimilation, melt hybridisation, and mantle metasomatic reactions from the laboratory experiments that have been carried out. First, olivine is consumed in these reactions (i.e., reactant phases are olivine and pristine adakite melt), and the primary metasomatic phases are orthopyroxene and amphibole or phlogopite (i.e., products). Secondly, at high melt:rock ratios and/or relatively high temperatures, most, if not all, of the original olivine in the peridotite is consumed, and one is left with a hybridised adakitic melt in equilibrium with a pyroxenite reaction residue (Rapp et al. 1999; Prouteau et al. 2001). At low melt:rock ratios and/or lower temperatures, the metasomatic phase assemblage is dominated by amphibole/ phlogopite and orthopyroxene, and although generally some of the original olivine is also present, it is often not clear from the experiments whether or not ‘armouring’ of the original olivine by orthopyroxene has taken place (Sen & Dunn 1994; Rapp et al. 1999; Prouteau et al. 2001), In this case the trace element signature of the slab melt is eﬀectively transferred to the mantle wedge, and this adakite-metasomatised pyroxenite then becomes a potential source for ‘mantle-derived’ melts in the future. Martin et al. (2005) have classiﬁed the melts that are associated with these two diﬀerent outcomes of slab melt-peridotite reactions in the mantle wedge as ‘high-SiO2 adakites’ (HSA; mantle-hybridised adakites), and ‘low-SiO2 adakites’ (LSA; partial melts of adakite-metasomatised peridotite). It is important to note that, although the experimental analogues of ‘HSAs’ have been produced in the laboratory (e.g., Rapp et al. 1999), experimental ‘LSAs’ have not. The

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION

23

Figure 8 Schematic diagram summarising the possible modes of interactions between TTG liquids and mantle wedge peridotites. This diagram reports the temperature (T() vs. the bulk composition of the system (X). X ranges between a pure peridotite and a pure TTG liquid. In a ‘true’ pseudosection, both the liquidus and solidus curves would probably be stepped; however for purpose of simpliﬁcation, they are drawn as lines. The dotted line represents the (conceptual) limit of stability for olivine, i.e. the limit between peridotitic and pyroxenitic solids (residuum if in equilibrium with the melt, ‘metasomatised mantle’ otherwise). Sanukitoid liquids exist for bulk compositions that are mixtures of TTG and peridotite, and for melt fractions ranging between the solidus and the liquidus (white parallelogram ﬁeld). This ﬁeld can be reached along diﬀerent paths. Path (a) is the ‘one-stage’ process mentioned in the text. In this case the TTG liquid reacts with peridotites to give a sanukitoid liquid in equilibrium with a maﬁc or ultramaﬁc solid. HSA liquids probably form along this trend (for high melt:rock ratios); lower TTG proportions would rather result in LSA or sanukitoids. Path (b) is the ‘two-stage’ process, (b1) is the complete assimilation of TTG liquids via metasomatic reactions with the peridotite, and (b2) corresponds to subsequent melting of this modiﬁed composition. The black F/a arrow shows how the F/a ratio (discussed in section 3) increases, inside the inter solidus–liquidus domain.

present authors expect that in fact a petrogenetic and compositional continuum probably exists between HSA and LSA magmas, and where a given magma falls along that continuum is probably highly dependent upon the mechanism and dynamics of melt inﬁltration and propagation. The distinction between the two groups (HSAs and LSAs) may in fact lie in the nature of the mineral assemblage with which they may have equilibrated: experimental results have shown that HSA melts are in equilibrium with a pyroxenite residue (garnetamphibole/phlogopite), whereas LSA melts are likely to be in equilibrium with an olivine-bearing (i.e., peridotite or garnet peridotite) phase assemblage, with the melting reaction being controlled by the breakdown of metasomatic amphibole and/or phlogopite. Hirose (1997) does report partial melts of hydrous lherzolite at 1·0 GPa that have the appropriate major-element composition of LSAs (e.g., SiO2 =54–60 wt.%), and are in equilibrium with a spinel peridotite residue (ol+opxcpx+spl). Unfortunately, in this paper no trace element data was available. In the present

interpretation of LSA and sanukitoid petrogenesis, the peridotitic source in these experiments would have experienced a metasomatism event prior to melting, in order for the TTG or ‘slab melt’ signature to be transferred to the mantle. The Late-Archaean sanukitoids are equated (Fig. 9) with LSA melts, possibly in equilibrium with an olivine-bearing pyroxenitic residue.

3. Discussion 3.1. Numerical modelling of sanukitoid genesis and diversity The chemical composition of sanukitoid liquids obtained by experimental fusion can be reproduced using numerical modelling, which provides important clues in understanding the details of petrogenetic mechanisms. For instance, it allows the processes that lead to the diﬀerence between low- and high-Ti

24

HERVE MARTIN ET AL.

concentration of the element in the TTG melt. Cmm is given by the mixing equation (1) Cmm =a.CTTG +(1a).Cm (Equation 1) The melt:rock ratio in the sense of Rapp et al. (1999, 2006), i.e. the relative melt and rock proportions actually involved in the interactions, is represented by a/1a. The composition of the melt (Cl) is given by the equilibrium melting equation (Shaw 1970) where F is the degree of melting and D the general distribution coeﬃcient: Cl ⫽

Figure 9 Diagrams comparing the composition of low-Ti (ﬁlled circles) and high-Ti (open squares) sanukitoids with SiO2 <62%, with liquids experimentally produced by interaction between a mantle peridotite and an Archaean TTG melt (grey stars) (Rapp et al. 2006): (A) Cr vs. MgO; (B) Sr/Y vs. La/Yb; (C) Nb/Y vs. La/Yb. Experimental melts display compositions identical with sanukitoids; in addition, inset (C) also shows that all experimental liquids plot in the light grey ﬁeld, which is the ﬁeld of modern high-Mg andesites considered as generated by melting of a mantle peridotite enriched by felsic melts (see Fig. 7 and text for more details).

sanukitoids to be addressed. Both types share many characteristics (Figs 2–6). However, systematic diﬀerences are observed for a few key geochemical parameters (mostly compatible elements ratios), as summarised in section 1.2, where it has been demonstrated that they reﬂect diﬀerent, albeit related, primitive magmas. These diﬀerences could be accounted for by two distinct processes: (1) diﬀerent degree of melting of a same source; and (2) diﬀerent degrees of metasomatism of a mantle peridotite by TTG melts. A simple model has been built (Moyen 2009) in order to take into account the co-variations of these two parameters; it mimics the experimental procedure and is simply based on the batch melting of a mantle peridotite, which composition has been modiﬁed by mixing with TTG melt. If the concentration of an element in the mantle before metasomatism is Cm, it becomes Cmm after metasomatism by a mass fraction a of TTG melt; CTTG being the

Cmm (Equation 2) F ⫹ D共1 ⫺ F兲

Partition coeﬃcients are from Rollinson (1993). For Fe and Mg, ‘pseudo’ partition coeﬃcients were calculated from minerals phases found in adakite-metasomatised peridotite nodules (Kepezhinskas et al. 1996; Gre´goire et al. 2008). Equilibrium melting is used to ensure consistency with experimental results, which are by construction at (or near) equilibrium. While simple, this model nevertheless allows for melt consumption (a>F) or, conversely, additions to the melt (F>a) during the interactions; the parameter F/a indicates the amount of melt gained or lost during the interactions. The diﬃculty consists in assigning realistic values for a and F, thus leading to a poorly constrained model. However, it appears that the only relevant parameter controlling the composition of the melt after the interaction with peridotite is the F/a ratio (Moyen 2009), i.e. the net gain or loss of melt during the process (Fig. 10). With increasing F/a values (i.e., towards a net melt gain – the interactions result in more addition to the melt than formation of new minerals by reactions between the melt and the peridotite), the resulting melt evolves to higher Mg#, but also lower incompatible element contents (REE, Sr, Y), and lower La/Yb and Sr/Y ratios. In contrast, low F/a values (i.e., the interactions result mostly in the formation of new minerals out of the melt, reducing its net volume) concentrate the incompatible elements, increase the La/Yb and Sr/Y ratios, and generate lower Mg# melts – although still with values higher than in the non-reacted TTG. In the same way as experimental petrology, this model is based on equilibrium between liquid and mineral phases for a given bulk composition X (P and T are implicit in the choice of mineral phases used for the calculation of D); consequently, the model cannot per se provide constrains on the P–T path (i.e., a one- or two-stage process) followed to reach the ﬁnal P–T–X state. The FeO, MgO and TiO2 behaviour is adequately accounted for by this model (Fig. 11). Indeed, decreasing F/a ratios result in a diminution of MgO correlated with a slight increase in FeO and a signiﬁcant augmentation of TiO2, thus adequately providing an explanation for the diﬀerence between low- and high-Ti sanukitoids. However, some of the high-Ti sanukitoids display TiO2 contents that are too high to be accounted by the present model. This can be explained by the choice of some parameters in the calculation: average TTG compositions (TiO2 =0·38%) were used as a ‘starting material’, but TTG compositions are actually more diverse, and TiO2 contents of 0·7–0·8% are not rare. Such a TiO2-richer TTG melt would easily account for the high-Ti sanukitoid characteristics. An alternative possibility is that the basic hypothesis of this model – equilibrium melting – is not true. Indeed, in the case of selective melting of only the Ti-rich portion of the metasomatised mantle (i.e., the amphibole-rich portion), without allowing bulk equilibrium with the residuum, the resulting liquid can be Ti-enriched well over that predicted by equilibrium models.

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION

25

Figure 10 Log(La), log(Yb), log(Sr) and Mg# vs. Log(F/a). F represents the proportion of liquid (degree of melting) and a the degree of metasomatism of a peridotitic source by TTG melts. The light grey domain corresponds to the F/a ratios that yield values closer to those measured in sanukitoids. Triangles and squares respectively represent the presence and absence of garnet during melting of the mantle peridotite. The thick black horizontal line corresponds to the average TTG composition used in the model. The D values used in computation are also given (see text and Moyen 2009 for more details).

Therefore, it is proposed that the diﬀerence between the high- and the low-Ti series essentially reﬂects diﬀerences in the F/a ratio (Fig. 8): the high-Ti group formed by low F/a (low degree of melting and/or more enriched source), and melting was probably far from equilibrium, allowing amphibole breakdown to control the melt’s composition without re-equilibrating with the rest of the rock. In contrast, the low-Ti group could correspond to higher (F/a) (high degree of melting and/or lower source enrichment), in conditions close to equilibrium melting. These two situations may reﬂect diﬀerent geological/geodynamical scenarios; indeed, whereas the low-Ti situation can arise in one single event, in which the mantle contamination and the generation of the sanukitoid melt are concomitant, the disequilibrium scenario envisioned for the high-Ti group requires a two-step process, in which amphibole is allowed to physically form, before being destroyed by a subsequent selective melting event. This requires a more complex geological history. However, Figure 10 shows that for all elements, the composition of both low- and high-Ti sanuki-

toids is achieved only for F/a<1, which means that the reaction between TTG melt and mantle is mostly TTG melt consuming leading to the formation of new minerals, in the mantle peridotite. This is consistent with the observed consumption of melt in assimilation experiments between depleted peridotite and ‘slab-derived’ TTG melts (Rapp et al. 1999). Moreover, in the case of a two-stage process, the amount of melt formed is lower than the melt consumed by the initial interactions.

3.2. Residual garnet and conditions of melting Sanukitoids have an average Yb content of 1·60 ppm, with a correlated high La/Yb=46·2. These values are intermediate between average TTG (Yb=0·64 ppm and La/Yb=59) and typical arc dacites and granodiorites (Yb=4·4 ppm and La/ Yb=9·7). Both experimental work (Rapp et al. 1999, 2006) and geochemical modelling done on sanukitoids as well as on low-silica adakites (LSA) (Moyen 2009) show that residual garnet is not absolutely necessary to account for the geochemical characteristics of most sanukitoids. The general shape of

26

HERVE MARTIN ET AL.

Figure 11 FeO vs. MgO and TiO2 vs. MgO diagrams comparing the results of geochemical modelling with the composition of sanukitoids. The dark grey ﬁeld is that of high-Ti sanukitoids whereas the light grey ﬁeld represents the low-Ti sanukitoids. In both models, the black and grey lines correspond to the evolution of liquid controlled by a and F parameters respectively. Dotted lines represent a constant F/a ratio. These diagrams show a change in the F/a ratio that accounts for the observed diversity of sanukitoids. (More details are given in the text).

the REE patterns of TTG seems to be transferred to sanukitoids, after a shift to higher values (Yb=0·64 ppm in TTG and 1·6 ppm in sanukitoids). However, experiments and modelling show that the presence of small amounts of residual garnet would make the acquisition of sanukitoid melt characteristics easier. Indeed, Figure 12 is a La vs. Yb plot for all sanukitoids; it points to diversity among sanukitoids. The range of La/Yb ratios observed is consistent with the presence of variable proportions of residual garnet during melting. As proposed by Francis & Ludden (1995) and Dalpe´ & Baker (2000), this garnet could be the result of peritectic reactions associated with the breakdown of amphibole. The variability in residual garnet could indicate that the TTG melt/mantle peridotite interactions were able to take place over a wide range of pressure, whatever the real mechanism is: direct hybridisation of the TTG melt or remelting of a peridotite metasomatised by TTG melts. It must also be noted that the La/Yb systematic does not coincide with the

low/high-Ti classiﬁcation, thus indicating that the two parameters are independent (both low and high F/a can be realised at shallow as well as at great depths). Garnet stability does not only depend on pressure, but is also controlled by rock composition. In a peridotite, garnet is stable only for pressure higher than 2·5 to 3·0 GPa; whereas for an amphibolitic (basaltic) composition it is stable at pressure as low as 1·0 to 1·2 GPa. Consequently, metasomatism of mantle peridotite by felsic melts will modify the peridotite composition: high TTG input resulting in garnet stable at lower pressure. Since the density of TTG melts is lower than that of mantle peridotite, they can only ascend through the mantle. This implies that the source for TTG is located below the mantle peridotite. In the case where garnet is stable in the mantle peridotite, this means that TTG melts are necessarily generated at depth greater than 40–60 km, depending on the amount of TTG melt having reacted with mantle peridotite. During melting of amphibolites, the garnet proportion increases with

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION

27

Figure 12 La vs. Yb diagram showing the diversity of La/Yb ratios in sanukitoids. La/Yb variations are not correlated with the two-fold low-Ti (ﬁlled circles) and high-Ti (open squares) sanukitoid classiﬁcation. The variability of La/Yb is accounted by diﬀerent proportions of residual garnet.

depth; in order to have enough garnet to produce a proper TTG composition, the depth of melting must be at least greater than 45–50 km (1·5 GPa) (Moyen & Stevens 2006; Nair & Chacko 2008): which is consistent with the depth of 40– 60 km deduced from sanukitoids. This requires a generally eﬃcient geodynamic process able to carry hydrated basalts to depths greater than 40–60km. The only mechanism able to straightforwardly and systematically realise such geometry with hydrous basalts under a mantle slice, at depths of >40–60 km, is subduction. Therefore, sanukitoids can be regarded as markers of at least some kind of subduction, with burial into the mantle of signiﬁcant portions of surface maﬁc hydrated material.

3.3. Geodynamic implications and Archaean–Proterozoic boundary The sanukitoids mainly emplaced at the Archaean–Proterozoic boundary, during a period where the dominant mechanism of genesis of the juvenile continental crust changed from melting of hydrous basalt (Archaean) to fusion of the ﬂuid metasomatised mantle peridotite (post-Archaean). As discussed earlier, evidence supports the genesis of most TTG and sanukitoids in a subduction environment; following this logic, the present authors can propose a general model for the generation of juvenile continental crust across the Archaean–Proterozoic boundary. Today, the subducted oceanic crust is old (60 Ma on average) and cold; it contributes to the cooling of the mantle wedge, such that geothermal gradient along the Benioﬀ plane is low (Fig. 13) and dehydration reactions in subducted basalt occur before it reaches its hydrous solidus. Consequently, the oceanic slab loses its water and is unable to melt at low temperature. Fluids liberated by dehydration reactions rise up towards the surface through the mantle wedge and induce its partial melting. The rising ﬂuids also transfer soluble elements such as LILE and LREE from the subducted lithosphere into the mantle wedge. In other words, most of the present-day juvenile continental crust is generated in subduction geodynamic environments, by the melting of a mantle wedge, whose composition has been modiﬁed by ﬂuids liberated by dehydration of the crustal portions of subducted slabs of oceanic lithosphere. During the Archaean, Earth heat produc-

tion was greater, resulting in higher geothermal gradients in subduction zones. Along such high geothermal gradients, the order in which the dehydration reaction and hydrous solidus curves are crossed is inverted (Fig. 13): the subducted oceanic slab reaches its hydrous solidus temperature before dehydration begins and is able to melt at relatively low temperature and shallow depth, but deep enough for garnet to be stable in the residue, giving rise to TTG magmas. Consequently, TTG is the result of relatively shallow depth melting of subducted oceanic crust metamorphosed into garnet amphibolite or eclogite. Today, in some exceptional cases, in subduction zones, where an abnormally young oceanic crust is subducted (active ridge subduction such as Patagonia), adakitic magmas are generated; which are very similar to Archaean TTG. This shows that hot geothermal gradients in subduction zones are able to result in slab melting and TTG-like magma genesis (see Martin 1999; Smithies 2000; Martin et al. 2005, for discussion and overviews). Recently, Smithies (2000), Smithies & Champion (2000) and Martin & Moyen (2002), showed that the chemical composition of the parental magma of TTG has changed through Archaean times. These authors consider that change as progressive, whereas Condie (2005) sees it as an abrupt event at about 2·7 Ga. According to Martin & Moyen (2002), the Mg# of the more primitive TTG magmas increased from maximum values of 0·45 at 4·0 Ga to 0·65 at 2·5 Ga. In the same period, the maximum concentrations of Ni and Cr increased, as Sr did, from w550 ppm at 3·8 Ga to w1200 ppm at 2·5 Ga (Fig. 14). On the other hand, the high Mg# (0·65) of the younger TTG (<3·0 Ga), is higher than values determined by experimental high-pressure melting of basaltic material, which requires that a mantle component played a role in the genesis of these magmas. Consequently, this time-related change is interpreted in terms of a secular increase of the interactions between the TTG parental and the mantle wedge peridotites (e.g. Maury et al. 1996; Rapp et al. 1999; Smithies 2000; Martin & Moyen 2002). Similarly, these authors interpret the Sr content in TTG parental magmas in terms of plagioclase stability; indeed, not only Sr content increases in course of time, but also (CaO+Na2O) and Al2O3. As plagioclase stability is pressure dependent; they conclude that, from 4·0 Ga to 2·5 Ga the depth of melting of the subducted slab progressively increased.

28

HERVE MARTIN ET AL.

Figure 13 P–T diagram and synthetic cross-sections of subduction zones (after Martin & Moyen 2002). (1) In the Early Archaean (4·0 Ga) the geothermal gradient along the Benioﬀ plane was very high, thus the subducted slab melts at shallow depth. Due to the thinness of the wedge and the low temperature, mantle and melt interactions are limited or absent; (2) After 3·0 Ga, the Earth was cooler, the geothermal gradient was lower and slab melting occurred at greater depth. The overlying mantle wedge is thick and hot, and interactions can occur between mantle and slab melts; (3) At the Archaean–Proterozoic transition, geothermal gradients are too low to allow a high degree of slab melting. Slab melts are almost totally consumed in a reaction with the mantle. Low-Ti sanukitoid are assumed to be the result of a single event of contamination of slab melts by peridotite, whereas low-Ti sanukitoids formed through a two-step process where the mantle metasomatised by slab melts is subsequently molten; (4) After 2·5 Ga, geothermal gradients are so low that slab melting is precluded. The oceanic crust dehydrates and the liberated ﬂuids metasomatise the mantle wedge, whose melting produces modern arc magmatism. Dehydration reactions: H=hornblende out; A=antigorite out; C=chlorite out; Ta=talc out; Tr=tremolite out; Z=zoisite out. G and P lines represent stability ﬁelds of garnet and plagioclase respectively. Grey ﬁeld is P–T TTG window. OC=oceanic crust; CC=continental crust; dotted line=solidus of hydrous mantle; black areas=magma; dotted area=ﬂuids; SMMP=slab melt metasomatised peridotite; FMP=ﬂuid metasomatised peridotite.

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION

29

Figure 14 Diagrams showing the time-evolution of MgO and Sr content of the primitive TTG parental magmas (grey circles) from 4·0 Ga to 2·5 Ga (after Martin & Moyen 2002). The less diﬀerent low-Ti (black circles) and high-Ti (open squares) sanukitoids have both MgO and Sr content far greater than in TTG, thus pointing to the fact that they were generated from a diﬀerent source or/and a diﬀerent mechanism.

However, most basalt melting experiments show that the amount of plagioclase stable together with garnet remains relatively low. Consequently, another possibility would be that in absence of signiﬁcant amounts of residual plagioclase, Sr has an incompatible behaviour, its content in the magmatic liquid becoming only dependent on the degree of melting. In this case, the augmentation of Sr content in TTG magmas in the course of time would only reﬂect a decrease in the degree of melting of the basaltic source of TTG; which would also be consistent with the progressive cooling of the Earth. However, lower degrees of melting should result in lower Mg# and Ni and Cr contents in TTG magma, thus leading to an anticorrelation between Sr and Mg# (and Ni and Cr), whereas the opposite is observed. Whatever the cause of the timedependent variation of Sr may be, it can be interpreted in terms of progressive cooling of the Earth, resulting in lower geothermal gradients in subduction systems (Fig. 13), thus leading to deeper slab melting. It must be noted that sanukitoids do not plot on this ‘trend’ but display signiﬁcantly higher values for MgO and Sr (Fig. 14) as well as for Ni and Cr. The present authors propose that throughout the Archaean, the Earth’s heat production was high enough (curves 1 and 2, Fig. 13) to allow a high degree of melting of the subducted slab, consequently the ‘eﬀective melt:rock ratio’ (as discussed above) was high and all slab melt was not consumed by mantle interaction, and was consequently emplaced largely as TTG. However, with time, Earth’s geothermal gradient decreased and slab melting took place at progressively greater depths, such that some degree of interaction with peridotitic lithologies of the mantle wedge became increasingly likely. At about 2·7–2·5 Ga ago, further reduction in geothermal gradients resulted in smaller degrees of melting of the subducted slab, such that most of the TTG melts coming oﬀ the slab would be consumed in metasomatic reactions with peridotite. In these conditions, the mantle imprint would be stronger and could be achieved in two diﬀerent ways: (1) in a single event where the mantle contamination and the generation of the sanukitoid melt are concomitant; this mechanism could result in low-Ti sanukitoid genesis; or (2) in a two-step process in which metasomatic minerals must crystallise in the mantle peridotite, before being destroyed by a subsequent melting event, thus giving rise to high-Ti sanukitoids (curve 3, Fig. 13). After Archaean times geothermal gradients were too low to allow

slab melting and modern continental crust was generated by ﬂuid metasomatised mantle wedge melting (curve 4, Fig. 13). This model accounts for the ‘transitory’ character of sanukitoids, as well as for their location at the Archaean–Proterozoic boundary.

4. Summary and conclusion The temporal evolution of magmas at the Archaean– Proterozoic boundary can be synthesised as follows (Fig. 15): 1. In the Early Archaean (T>3·5 Ga), the terrestrial heat production was important, such that the subducted basalts underwent high degree of melting at shallow depth. The ‘eﬀective melt:rock ratio’ in the mantle above the source of TTGs was high and the TTG magmas emplaced after minor or no interaction with the peridotite. 2. In the Middle to Late Archaean, the heat production was lower and the degree of melting of the subducted slab was slightly lower and took place at greater depth. However, the eﬃciency of slab-melting was high enough to maintain high slab-melt/mantle peridotite ratios. Thus, the slab-melt was not totally consumed in reaction with peridotite (Rapp et al. 1999, 2006), and consequently TTG magma was emplaced into the crust. 3. During the Late Archaean, and particularly at the Archaean–Proterozoic boundary, the Earth heat production and the eﬃciency of slab-melting had both declined. Slab-melt/mantle peridotite ratios had correspondingly declined such that slab-melts were almost totally consumed in reaction with mantle peridotite, thus producing Archaean sanukitoids. These later could result from either a single event where the mantle contamination and the generation of the sanukitoid melt are concomitant, (low-Ti sanukitoids) or from a two-step process where the mantle metasomatised by slab melts is subsequently molten (high-Ti sanukitoids). 4. Since the Lower Proterozoic, Earth’s heat production was too low to allow subducted slab-melting under ‘normal’ conditions. Consequently, a slab dehydrates and classic BADR calc-alkaline magmatism results from melting of a peridotite that has been metasomatised by slab dehydration ﬂuids.

30

HERVE MARTIN ET AL.

Figure 15 Schematic diagram illustrating the evolution of the juvenile crustal magmatism in course of the Earth’s history. The thickness of the dark domain is qualitatively indicative of the volumetric importance of the magmatism. It clearly points to a large domain of overlapping at the Archaean–Proterozoic transition (2·5 Ga).

In other words, with regards to continental crust genesis, it appears that the Archaean–Proterozoic transition corresponds to a major change from a crustal (oceanic basalts) to a mantle source. The sanukitoid magmatism that shows both crustal and mantle imprint appears to be a good marker of these changes. Of course, this presentation of the timing of juvenile continental crust genesis over the course of Earth history is very simplistic and schematic. If today, in subduction environments, BADR magmatism is the more widespread, slab melting can also locally take place, leading to adakite genesis. Whenever, Archaean-like thermal regimes are established in the modern Earth, TTG-like (adakite) magmas are produced. However, this mechanism remains very minor on a global scale, but was once dominant. During Earth cooling, new thermodynamic conditions appeared, but not everywhere at the same time, because local conditions play an important role in controlling these thermodynamic regimes. This results in an overlap of petrogenetic mechanisms during time (Fig. 15), leading, for instance, to simultaneous TTG and sanukitoid magmatism: the apparently progressive character of petrogenetic changes near the Archaean–Proterozoic boundary is a statistic eﬀect that reﬂects the progressive onset and development of some petrogenetic processes, and the diminishing importance of others. The dominant process of the Early Archaean was shallow slab melting that generated low-Mg TTG; this process became progressively less important during the later periods, and was all but replaced during Late Archaean times by deeper slab melting, which allowed for more signiﬁcant interactions with mantle peridotite, and resulted in increased production of high-Mg TTG. Low-Mg TTG persisted late into the Archaean, even if not as important as they once were. In turn, the importance of high-Mg TTG decreased during the Proterozoic, and became a very rare process in the Phanerozoic; it was progressively replaced by BADR magmatic activity, which ﬁrst appears in the Early Proterozoic and which became progressively dominant.

Sanukitoids represent the ‘missing link’ between the (high-Mg) TTGs and the BADR; indeed, their transitional character is not only compositional but also chronologic, since they were emplaced about 2·5 Ga ago, during the transitional period between two epochs respectively dominated by TTG and BADR juvenile crustal magmatism.

5. References Albare`de, F. 1998. The growth of continental crust. Tectonophysics 296, 1–14. Althoﬀ, F. J. 1996. Etude pe´trologique et structurale des granitoı¨des de Marajoara (Para´, Bre´sil): leur roˆle dans l’e´volution arche´enne du craton amazonien (2·7–3·2 Ga). Nancy: Henri Poincarre´. 296 pp. Arth, J. G. & Hanson, G. N. 1972. Quartz diorites derived by partial melting of eclogite or amphibolite at mantle depth. Contributions to Mineralogy and Petrology 37, 164–74. Atherton, M. P. & Petford, N. 1993. Generation of sodium-rich magmas from newly underplated basaltic crust. Nature 362, 144–6. Balakrishnan, S. & Rajamani, V. 1987. Geochemistry and petrogenesis of granitoids around Kolar schist belt: constraints for crustal evolution in Kolar area. Journal of Geology 95, 219–40. Barker, F. & Arth, J. G. 1976. Generation of trondhjemitic–tonalitic liquids and Archaean bimodal trondhjemite–basalt suites. Geology 4, 596–600. Barth, M. G., Foley, S. F. & Horn, I. 2002. Partial melting in Archaean subduction zones: constraints from experimentally determined trace element partition coeﬃcients between eclogitic minerals and tonalitic melts under upper mantle conditions. Precambrian Research 113, 323–40. Barton, J. J. M., Doig, R., Smith, C. B., Bohlender, F. & van Reenen, D. D. 1992. Isotopic and REE characteristics of the intrusive charnoenderbite and enderbite geographically associated with the Matok Pluton, Limpopo Belt, southern Africa. Precambrian Research 55 (1–4), 451–67. Be´dard, J. 2006. A catalytic delamination-driven model for coupled genesis of Archaean crust and subcontinental lithospheric mantle. Geochimica et Cosmochimica Acta 70, 747–71. Be´dard, P. 1996. Archean High-Mg monzodiorite plutonic suite: A reevaluation of the parental magma and diﬀerentiation. Journal of Geology 104, 713–28.

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION Blichert-Toft, J. & Albare`de, F. 2008. Hafnium isotopes in Jack Hills zircons and the formation of the Hadean crust. Earth and Planetary Science Letters 265 (3–4), 686–702. Bourdon, E., Eissen, J.-P., Gutscher, M.-A., Monzier, M., Hall, M. L. & Cotten, J. 2003. Magmatic response to early aseismic ridge subduction: the Ecuadorian margin case (South America). Earth and Planetary Science Letters 205 (3–4), 123–38. Brown, G. C. 1985. Processes and problems in the continental lithosphe`re: geological history and physical implications. In Snelling, N. (ed.) The chronology of the geological record. Geological Society, London, Memoir 10, 326–34. Bath, UK: The Geological Society Publishing House. Calmus, T., Aguillon-Robles, A., Maury, R. C., Bellon, H., Benoit, M., Cotten, J., Bourgois, J. & Michaud, F. 2003. Spatial and temporal evolution of basalts and magnesian andesites (‘bajaites’) from Baja California, Mexico: the role of slab melts. Lithos 66 (1–2), 77–105. Cavosie, A. J., Wilde, S. A., Liu, D., Weiblen, P. W. & Valley, J. W. 2004. Internal zoning and U–Th–Pb chemistry of Jack Hills detrital zircons: a mineral record of early Archean to Mesoproterozoic (4348–1576 Ma) magmatism. Precambrian Research 135 (4), 251–79. Cavosie, A. J., Valley, J. W., Wilde, S. A. & E.I.M.F. 2005. Magmatic [delta]18O in 4400–3900 Ma detrital zircons: A record of the alteration and recycling of crust in the Early Archean. Earth and Planetary Science Letters 235 (3–4), 663–81. Cavosie, A. J., Valley, J. W. & Wilde, S. A. 2006. Correlated microanalysis of zircon: Trace element, [delta]18O, and U–Th–Pb isotopic constraints on the igneous origin of complex >3900 Ma detrital grains. Geochimica et Cosmochimica Acta 70 (22), 5601–16. Chardon, D., Peucat, J.-J., Jayananda, M., Choukroune, P. & Fanning, M. C. 2002. Archean granite-greenstone tectonics at Kolar (South India): Interplay of diapirism and bulk inhomogeneous contraction during juvenile magmatic accretion. Tectonics 21 (3), 7.1–7.17. Condie, K. C. 1981. Archaean greenstone belts. Amsterdam: Elsevier. 434 pp. Condie, K. C. 1989. Plate tectonics and crustal evolution. Oxford: Pergamon. 476 pp. Condie, K. C. 2005. TTGs and adakites: are they both slab melts? Lithos 80, 33–44. Dalpe´, C. & Baker, D. R. 2000. Experimental investigation of largeion-lithophile-element-, high-ﬁeld-strength-element- and rareearth-element-partitioning between calcic amphibole and basaltic melt: the eﬀects of pressure and oxygen fugacity. Contributions to Mineralogy and Petrology 140, 233–50. de Wit, M. J. 1998. On Archean granites, greenstones, cratons and tectonics: does the evidence demand a verdict? Precambrian Research 91 (1–2), 181–226. Foley, S. F., Tiepolo, M. & Vannucci, R. 2002. Growth of early continental crust controlled by melting of amphibolite in subduction zones. Nature 417, 637–40. Forneris, J. F. & Holloway, J. R. 2003. Phase equilibria in subducting basaltic crust: implications for H2O release from the slab. Earth and Planetary Science Letters 214, 187–201. Francis, D. & Ludden, J. 1995. The signature of amphibole in maﬁc alkaline lavas, a study in the Northern Canadian Cordillera. Journal of Petrology 36, 1171–91. Frost, C. D., Frost, B. R., Chamberlain, K. R. & Hulsebosch, T. P. 1998. The late Archaean history of the Wyoming province as recorded by granitic magmatism in the Wind River Range, Wyoming. Precambrian Research 89, 145–73. Glikson, A. Y. 1971. Primitive Archaean element distribution patterns: chemical evidence and geotectonic signiﬁcance. Earth and Planetary Science Letters 12, 309–20. Gre´goire, M., Je´go, S., Maury, R. C., Polve´, M., Payot, B., Tamayo, R. A. Jr. & Yumel, G. P. Jr. 2008. Metasomatic interactions between slab-derived melts and depleted mantle: Insights from xenoliths within Monglo adakite (Luzon arc, Philippines). Lithos 103 (3–4), 415–30. Halla, J. 2005. Late Archean high-Mg granitoids (sanukitoids) in the southern Karelian domain, eastern Finland: Pb and Nd isotopic constraints on crust-mantle interactions. Lithos 79 (1–2), 161–78. Harrison, T. M. & Schmitt, A. K. 2007. High sensitivity mapping of Ti distributions in Hadean zircons. Earth and Planetary Science Letters 261(1–2), 9–19. Hirose, K. 1997. Melting experiments on lherzolite KLB-1 under hydrous conditions and the generation of high-magnesian andesitic melts. Geology 25, 42–4.

31

Hoﬀer, G. 2008. Fusion partielle d’un manteau me´tasomatise´ par un liquide adakitique: approches ge´ochimique et expe´rimentale de la gene`se et de l’e´volution de l’arrie`re arc e´quatorien. PhD Thesis, Universite´ Blaise Pascal, Clermont Ferrand, France. 320 pp. Jahn, B. M., Glikson, A. Y., Peucat, J.-J. & Hickman, A. H. 1981. REE geochemistry and isotopic data of Archaean silicic volcanics and granitoids from the Pilbara Block, western Australia: implications for the early crustal evolution. Geochimica et Cosmochimica Acta 45, 1633–52. Jahn, B.-M., Gruau, G., Capdevila, R., Cornichet, J., Nemchin, A., Pidgeon, R. & Rudnik, V. A. 1998. Archaean crustal evolution of the Aldan shield, Siberia: geochemical and isotopic constraints. Precambrian Research 91, 333–63. Jayananda, M., Martin, H., Peucat, J.-J. & Mahabaleswar, B. 1995. Late Archaean crust-mantle interactions in the Closepet granite, Southern India: evidence from Sr–Nd isotopes, major and trace element geochemistry. Contributions to Mineralogy and Petrology 119, 314–29. Johnson, A. D. & Wyllie, P. J. 1989. The system tonalite-peridotiteH2O at 30 kbar, with applications to hybridization in subduction zone magmatism. Contributions to Mineralogy and Petrology 102, 257–64. Kamber, B. S., Ewart, A., Collerson, K. D., Bruce, M. C. & McDonald, G. A. 2002. Fluid-mobile trace element constraints on the role of slab melting and implications for Archaean crustal growth models. Contributions to Mineralogy and Petrology 144, 38–56. Ka¨pyaho, A. 2006. Whole-rock geochemistry of some tonalite and high Mg/Fe gabbro, diorite, and granodiorite plutons (sanukitoid suites) in the Kuhmo district, eastern Finland. Bulletin of the Geological Society of Finland 78, 121–41. Kelemen, P. B., Shimizu, H. & Dunn, T. 1993. Relative depletion of niobium in some arc magmas and the continental crust: partitioning of K, Nb, La and Ce during melt/rock reaction in the upper mantle. Earth and Planetary Science Letters 120, 111–34. Kelemen, P. B., Hart, S. R. & Bernstein, S. 1998. Silica enrichment in the continental upper mantle via melt/rock reaction. Earth and Planetary Science Letters 164, 387–406. Kepezhinskas, P. K., Defant, M. J. & Drummond, M. S. 1996. Progressive enrichment of island arc mantle by melt-peridotite interaction inferred from Kamchatka xenoliths. Geochimica et Cosmochimica Acta 60, 1217–29. Kleinhanns, I. C., Kramers, J. D. & Kamber, B. S. 2003. Importance of water for Archaean granitoid petrology: a comparative study of TTG and potassic granitoids from Barberton Mountain and, South Africa. Contributions to Mineralogy and Petrology 145, 377–89. Kovalenko, A., Clemens, J. D. & Savatenkov, V. 2005. Petrogenetic constraints for the genesis of Archaean sanukitoid suites: geochemistry and isotopic evidence from Karelia, Baltic Shield. Lithos 79 (1–2), 147–60. Kreissig, K., Holzer, L., Frei, R., Villa, I. M., Kramers, J. D., Kroner, A., Smit, C. A. & van Reemen, D. D. 2001. Geochronology of the Hout River Shear Zone and the metamorphism in the Southern Marginal Zone of the Limpopo belt, Southern Africa. Precambrian Research 109, 145–73. Krogstad, E. J., Hanson, G. N. & Rajamani, V. 1991. U–Pb ages of zircon and sphene for two gneiss terranes adjacent to the Kolar schist belt, South India: evidence for separate crustal evolution histories. Journal of Geology 99, 801–16. Krogstad, E. J., Hanson, G. N. & Rajamani, V. 1995. Sources of continental magmatism adjacent to late Archaean Kolar suture zone, south India: distinct isotopic and elemental signatures of two late Archaean magmatic series. Contributions to Mineralogy and Petrology 122, 159–73. Leite, A. A. S., Dall’Agnol, R., Macambira, M. J. B. & Althoﬀ, F. J. 2004. Geologia e geocronologia dos granito´ides arqueanos da regia˜o de Xinguara e suas implicac¸o˜es na evoluc¸a˜o do Terreno Granito–Greenstone de Rio Maria, Cra´ton Amazoˆnico. Revista Brasileira de Geociencias 34 (4), 447–58. Liu, J., Bohlen, S. R. & Ernst, W. G. 1996. Stability of hydrous phases in subducting oceanic crust. Earth and Planetary Science Letters 143, 167–71. Lobach-Zhuchenko, S. B., Kovalenko, A. V., Krylov, I. N., Levskii, L. K. & Bogomolov, E. S. 2000. Geochemistry and petrology of the ancient Vygozero granitoids, Southern Karelia. Geochemistry International 38, 584–99. Lobach-Zhuchenko, S. B., Rollinson, H. R., Chekulaev, V. P., Arestovaa, N. A., Kovalenko, A. V., Ivanikov, V. V., Guseva, N. S., Sergeev, S. A., Matukov, D. I. & Jarvis, K. E. 2005. The Archaean sanukitoid series of the Baltic Shield: geological setting,

32

HERVE MARTIN ET AL.

geochemical characteristics and implications for their origin. Lithos 79 (1–2), 107–28. Lobach-Zhuchenko, S. B., Rollinson, H., Chekulaev, V. P., Savatenkov, V. M., Kovalenko, H., Martin, A. V., Guseva, N. S. & Arestova, N. A. 2008. Petrology of a Late Archaean, Highly Potassic, Sanukitoid Pluton from the Baltic Shield: Insights into Late Archaean Mantle Metasomatism. Journal of Petrology 49 (3), 393–420. Macpherson, C. G., Dreher, S. T. & Thirlwall, M. F. 2006. Adakites without slab melting: High pressure diﬀerentiation of island arc magma, Mindanao, the Philippines. Earth and Planetary Science Letters 243 (3–4), 581–93. Martin, E., Martin, H. & Sigmarsson, O. 2008. Could Iceland be a modern analogue for the Earth’s early continental crust? Terra Nova 20, 463–68. Martin, E. & Sigmarsson, O. 2007. Low-pressure diﬀerentiation of tholeiitic lavas as recorded in segregation veins from Reykjanes (Iceland), Lanzarote (Canary Islands) and Masaya (Nicaragua). Contributions to Mineralogy and Petrology 154, 559–73. Martin, H. 1986. Eﬀect of steeper Archean geothermal gradient on geochemistry of subduction-zone magmas. Geology 14, 753–6. Martin, H. 1987. Petrogenesis of Archaean trondhjemites, tonalites and granodiorites from eastern Finland: major and trace element geochemistry. Journal of Petrology 28(5), 921–53. Martin, H. 1993. The mechanisms of petrogenesis of the Archaean continental crust – comparison with modern processes. Lithos 30, 373–88. Martin, H. 1994. The Archean grey gneisses and the genesis of the continental crust. In Condie, K. C. (ed.) The Archean crustal evolution. Developments in Precambrian Geology, 205–59 Amsterdam: Elsevier. Martin, H. 1999. The adakitic magmas: modern analogues of Archaean granitoids. Lithos 46(3), 411–29. Martin, H., Chauvel, C. & Jahn, B. M. 1983. Major and trace element geochemistry and crustal evolution of granodioritic Archaean rocks from eastern Finland. Precambrian Research 21, 159–80. Martin, H., Sabate´, P., Peucat, J.-J. and Cunha, J. C. 1997. Crustal evolution in the early Archaean of South America: example of the Sete Voltas Massif (Bahia State, Brazil). Precambrian Research 82, 35–62. Martin, H., Smithies, R. H., Rapp, R., Moyen, J.-F. & Champion, D. 2005. An overview of adakite, tonalite–trondhjemite–granodiorite (TTG), and sanukitoid: relationships and some implications for crustal evolution. Lithos 79 (1–2), 1–24. Martin, H. & Moyen, J.-F. 2002. Secular changes in TTG composition as markers of the progressive cooling of the Earth. Geology 30(4), 319–22. Masuda, A., Nakamura, N. & Tanaka, T. 1973. Fine structures of mutually normalized rare-earth patterns of chondrites. Geochimica et Cosmochimica Acta 37, 239–44. Maury, R. C., Sajona, F. G., Pubellier, M., Bellon, H. & Defant, M. J. 1996. Fusion de la crouˆte oce´anique dans les zones de subduction/ collision re´centes: l’exemple de Mindanao (Philippines). Bulletin de la Socie´te´ Ge´ologique de France 167 (5), 579–95. McGregor, V. R. 1979. Archaean grey gneisses and the origin of the continental crust: evidence the Godthaab region, west Greenland. In Barker, F. (ed.) Trondhjemites, dacites and related rocks, 169–204. Amsterdam: Elsevier. Medeiros, H. & Dall’Agnol, R. 1988. Petrologia da porc¸a˜o leste do Bato´lito Granodiorı´tico Rio Maria, sudeste do Para´. In Anais do 35 Congresso Brasileiro de Geologia, 1488–99. Bele´m: Sociedade Brasileira de Geologia. Millonig, L., Zeh, A., Gerdes, A. & Klemd, R. 2008. Neoarchaean high-grade metamorphism in the Central Zone of the Limpopo Belt (South Africa): Combined petrological and geochronological evidence from the Bulai pluton. Lithos 103 (3–4), 333–51. Monzier, M., Robin, C., Hall, M. L., Cotton, J., Mothes, P., Eissen, J.-P. & Samaniego, P. 1997. Les adakites d’Equateur: mode`le pre´liminaire. Compte Rendus de l’Acade´mie des Sciences de Paris 324, 545–52. Moyen, J.-F. 2009. High Sr/Y and La/Yb ratios: the meaning of the ‘adakitic signature’. Lithos 112(3–4) 556–74. Moyen, J.-F., Martin, H. & Jayananda, M. 1997. Origine du granite ﬁni-Arche´en de Closepet (Inde du Sud): apports de la mode´lisation ge´ochimique du comportement des e´le´ments en traces. Compte Rendus de l’Acade´mie des Sciences de Paris 325, 659–64. Moyen, J.-F., Martin, H. & Jayananda, M. 2001a. Multi-element geochemical modelling of crust-mantle interactions during lateArchaean crustal growth: the Closepet granite (South India). Precambrian Research 112, 87–105.

Moyen, J.-F., Ne´de´lec, A., Martin, H. & Jayananda, M. 2001b. Contrasted granite emplacement modes within an oblique crustal section: the Closepet Granite, South India. Physics and Chemistry of the Earth (A) 26 (4–5), 295–301. Moyen, J.-F., Martin, H., Jayananda, M. & Auvray, B. 2003. Late Archaean granites: a typology based on the Dharwar Craton (India). Precambrian Research 127 (1–3), 103–23. Moyen, J.-F. & Stevens, G. 2006. Experimental constraints on TTG petrogenesis: implications for Archaean geodynamics. In Benn, K. Condie, K. C. & Mareschal, J. C. (eds) Archaean geodynamics and environments. AGU Monograph 164, 149–75. Washington, DC: American Geophysical Union. Nair, R. & Chacko, T. 2008. Role of oceanic plateaus in the initiation of subduction and origin of continental crust. Geology 36 (7), 583–6. Nair, R. K. & Chacko, T. 2005. Oceanic plateaus: nuclei for Archean cratons. Geological Society of America Annual Meeting, Salt Lake City, 494. Boulder, Colorado and Lawrence, Kansas: The Geological Society of America and University of Kansas Press. Nicholls, I. A. & Ringwood, A. E. 1973. Production of silica saturated tholeiitic magmas in island arcs. Earth and Planetary Science Letters 16, 243–6. Pawley, A. R. & Holloway, J. R. 1993. Water source for subduction zone volcanism: new experimental constrains. Science 121, 664–7. Prouteau, G., Scaillet, B., Pichavant, M. & Maury, R. C. 2001. Evidence for mantle metasomatism by hydrous silicic melts derived from subducted oceanic crust. Nature 410, 197–200. Querre´, G. 1985. Palingene`se de la crouˆte continentale a` l’Arche´en: les graniteı¨des tardifs (2·5–2·4 Ga) de Finlande orientale; pe´trologie et ge´ochimie. Me´moires et Documents du Centre Armoricain d’Etude Structurale des Socles 2. Rennes: Universite´ de Rennes. Rapp, R. P., Watson, E. B. & Miller, C. F. 1991. Partial melting of amphibolite/eclogite and the origin of Archaean trondhjemites and tonalites. Precambrian Research 51, 1–25. Rapp, R. P., Shimizu, N., Norman, M. D. & Applegate, G. S. 1999. Reaction between slab-derived melts and peridotite in the mantle wedge: experimental constraints at 3·8 GPa. Chemical Geology 160, 335–56. Rapp, R. P., Shimizu, N. & Norman, M. D. 2003. Growth of early continental crust by partial melting of eclogite. Nature 425, 605–9. Rapp, R. P., Laporte, D., Martin, H. & Shimizu, N. 2006. Experimental insights into slab-mantle interactions in subduction zones: Melting of adakite-metasomatized peridotite and the origin of the ‘arc signature’. Geochimica et Cosmochimica Acta 70 (18, Supplement 1), A517. Rapp, R. P. & Watson, E. B. 1995. Dehydration melting of metabasalt at 8–32 kbar: implications for continental growth and crust-mantle recycling. Journal of Petrology 36 (4), 891–931. Reddy, G. S. 1991. Geochemistry and petrogenesis of granitic rocks around Sakarsanahalli (Kolar) South India. PhD Thesis, Bangalore University, Bangalore. 147 pp. Rogers, N. W. & Setterﬁeld, T. N. 1994. Potassium and incompatibleelement enrichment in shoshonitic lavas from the Tavua volcano. Fiji. Chemical Geology 118, 43–62. Rollinson, H. 1993. Using geochemical data: evaluation, presentation, interpretation. London: Longman. 352 pp. Rollinson, H. 1997. Eclogite xenoliths in west African kimberlites as residues from Archaean granitoid crust formation. Nature 389, 173–6. Rudnick, R. L. 1995. Making continental crust. Nature 378, 571–7. Samsonov, A. V., Bogina, M. M., Bibikova, E. V., Petrova, A. Y. & Shchipansky, A. A. 2005. The relationship between adakitic, calc-alkaline volcanic rocks and TTGs: implications for the tectonic setting of the Karelian greenstone belts, Baltic Shield. Lithos 79 (1–2), 83–106. Sarvothaman, H. 2001. Archaean High-Mg granitoids of mantle origin in the Eastern Dharwar craton of Andhra Pradesh. Journal of the Geological Society of India 58, 261–8. Schmidt, M. W. & Poli, S. 1998. Experimentally based water budgets for dehydrating slabs and consequences for arc magma generation. Earth and Planetary Science Letters 163 (1–4), 361–79. Sekine, T. & Wyllie, P. J. 1982. The system granite-peridotite-H2O at 30 kbar, with applications to subduction zone magmatism. Contributions to Mineralogy and Petrology 81, 190–202. Sen, C. & Dunn, T. 1994. Experimental modal metasomatism of a spinel lherzolite and the production of amphibole-bearing peridotite. Contributions to Mineralogy and Petrology 119, 422–32. Shaw, D. M. 1970. Trace element fractionation during anatexis. Geochimica et Cosmochimica Acta 34, 237–43.

SANUKITOID MAGMATISM AT THE ARCHAEAN–PROTEROZOIC TRANSITION Shirey, S. B. & Hanson, G. N. 1984. Mantle derived Archaean monzodiorites and trachyandesites. Nature 310, 222–4. Shirey, S. B. & Hanson, G. N. 1986. Mantle heterogeneity and crustal recycling in Archaean granite-greenstone belts: evidence from Nd isotopes and trace elements in the Rainy Lake province, Ontario, Canada. Geochimica et Cosmochimica Acta 50, 2631–51. Smithies, R. H. 2000. The Archaean tonalite–trondhjemite– granodiorite (TTG) series is not an analogue of Cenozoic adakite. Earth and Planetary Science Letters 182, 115–25. Smithies, R. H., Van Kranendonk, M. J. & Champion, D. C. 2005. It started with a plume – early Archaean basaltic proto-continental crust. Earth and Planetary Science Letters 238 (3–4), 284–97. Smithies, R. H. & Champion, D. C. 1999. High-Mg diorite from the Archaean Pilbara Craton; anorogenic magmas derived from a subduction-modiﬁed mantle. Geological Survey of Western Australia, Annual Review 1998–99, 45–59. Smithies, R. H. & Champion, D. C. 2000. The Archaean high-Mg diorite suite: Links to Tonalite–Trondhjemite–Granodiorite magmatism and implications for early Archaean crustal growth. Journal of Petrology 41 (12), 1653–71. Stern, R. 1989. Petrogenesis of the Archaean sanukitoid suite. New York: State University at Stony Brook. 275 pp. Stern, R. A. & Hanson, G. N. 1991. Archaean high-Mg granodiorite: a derivative of Light Rare Earth enriched monzodiorite of mantle origin. Journal of Petrology 32, 201–38. Stevenson, R., Henry, P. & Garie´py, C. 1999. Assimilation-fractional crystallization origin of Archaean sanukitoid suites: Western Superior Province, Canada. Precambrian Research 96, 83–99. Sun, S. S. & McDonough, W. F. 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle compos-

33

ition and processes. In Saunders, S. D. & Norry, M. J. (eds) Magmatism in the ocean basins. Geological Society, London, Special Publication 42, 313–45. Bath, UK: The Geological Society Publishing House. Tatsumi, Y. 1989. Migration of ﬂuid phase and genesis of basalt magma in subduction zones. Journal of Geophysical Research 94, 4697–707. Tatsumi, Y. & Ishizaka, K. 1982. Origin of high-magnesian andesites in the Setouchi volcanic belt, southwest Japan, I. Petrographical and chemical characteristics. Earth and Planetary Science Letters 60 (2), 293–304. Viljoen, M. J. & Viljoen, R. P. 1969. The chemical evolution of the granitic rocks of the Barberton region. Geological Society of South Africa Special Publication 2, 189–220. Watson, E. B. & Harrison, T. M. 2005. Zircon thermometer reveals minimum melting conditions on earliest Earth. Science 308, 841–4. Wilde, S. A., Valley, J. W., Peck, W. H. & Graham, C. M. 2001. Evidence from detrital zircons for the existence of continental crust and oceans on the Earth 4·4 Ga ago. Nature 409, 175–78. Windley, B. F. & Bridgwater, D. 1971. The evolution of Archaean low- and high-grade terrains. Geological Society of Australia Special Publication 3, 33–46. Yaxley, G. M. & Green, D. H. 1998. Reactions between eclogite and peridotite: mantle refertilisation by subduction of oceanic crust. Schweizeriche Mineralogische und Petrographische Mitteilungen 78, 243–55. Zamora, D. 2000. Fusion de la crouˆte oce´anique subducte´e: approche expe´rimentale et ge´ochimique. PhD Thesis, Universite´ Blaise Pascal, Clermont-Ferrand, France. 314 pp.

MS received 1 February 2008. Accepted for publication 19 November 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 35–50, 2010 (for 2009)

The geochemistry of Archaean plagioclase-rich granites as a marker of source enrichment and depth of melting J.-F. Moyen1, D. Champion2 and R. H. Smithies3 1

Department of Geology, Geography and Environmental Studies, University of Stellenbosch, PVT Bag X1, 7602 Matieland, South Africa Email: [email protected] Current address: De´partement de Ge´ologie, Universite´ Jean-Monnet & CNRS 23, rue du Docteur Michelon, 42023 Saint-Etienne cedex, France Email: [email protected]

2

Geoscience Australia, Canberra, Australia Email: [email protected]

3

GSWA, Perth, W. Australia Email: [email protected]

ABSTRACT: In geochemical diagrams, granitoids deﬁne ‘trends’ that reﬂect increasing diﬀerentiation or melting degree. The position of an individual sample in such a trend, whilst linked to the temperature of equilibration, is diﬃcult to interpret. On the other hand, the positions of the trends within the geochemical space (and not the position of a sample within a trend) carry important genetic information, as they reﬂect the nature of the source (degree of enrichment) and the depth of melting. This paper discusses the interpretation of geochemical trends, to extract information relating to the sources of granitoid magmas and the depth of melting. Applying this approach to mid-Archaean granitoids from both the Barberton granite–greenstone terrane (South Africa) and the Pilbara Craton (Australia) reveals two features. The ﬁrst is the diversity of the group generally referred to as ‘TTGs’ (tonalites, trondhjemites and granodiorites). These appear to be composed of at least three distinct sub-series, one resulting from deep melting of relatively depleted sources, the second from shallower melting of depleted sources, and the third from shallow melting of enriched sources. The second feature is the contrast between the (spatial as well as temporal) distributions and associations of the granites in both cratons. KEY WORDS:

Geochemistry, geochemical diagrams, granitoids, trace elements

The bulk of the Archaean continental crust is made of granite or granitoids (e.g. Windley 1995), occurring either as welldeﬁned plutons, or as polyphase, deformed, commonly formerly partially melted ‘grey gneiss complexes’ (Martin 1994). Understanding their origin and petrogenesis is therefore critical to our understanding of the formation and diﬀerentiation of the Earth’s crust. Also, because they represent a strongly dominant rock type in the Archaean crust, granitoids can be useful geodynamic indicators, providing information can be extracted pertaining to the P–T regimes, the geometry of the crust, or the tectonic conditions in the crust during melting of their parental rocks, or the intrusion of granitic magmas. Classical thermobarometry, such as is used on metamorphic assemblages, is of limited use in granitoids, because such rocks typically lack ‘good’ metamorphic minerals such as garnet or aluminium silicates. However, it is increasingly recognised that the geochemistry of granites records at least some information not only on the source, but also of the condition of melting of this source (Moyen & Stevens 2006; Champion & Smithies 2007a; Collins 2007). This signal is obscured by the fact that even simple granitoid plutons typically display a range of composition, forming geochemical ‘trends’ from lower to higher silica

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016132

compositions. Whilst this evolution is customary referred to as a range of rocks, from ‘less diﬀerentiated’ to ‘more diﬀerentiated’, the trends can reﬂect either melt fractions (during melting), or latter diﬀerentiation (during cooling and emplacement). This paper discusses the role of petrogenetic eﬀects such as the degree and depth of melting, the nature of the source, and the amount of fractionation, on the shape and position of the trends in the composition space. It is shown that depth of melting and source composition exert a major control on the trends, and that the degree of melting or fractionation mostly results in composition changes within one given trend. Consequently, it is concluded that rather than individual analyses, the discussion on granites geochemistry should be based on trends aﬀecting whole rock units, as these carry the most valuable geodynamic information, and a graphical technique is proposed that simultaneously depicts information relating to source enrichment and to pressures of melting of the source. This analysis is then applied to the dominant type of Archaean granites – here referred to as ‘plagioclase-rich granitoids’ or ‘TTGM’ (deﬁned below). Since granitoids represent a large portion of the exposed Archean crust, and in some areas are the only remaining evidence for whole portions of the accretion history of some

36

J.-F. MOYEN ET AL.

Archaean blocks, this approach provides a very valuable tool in interpreting Archaean geodynamics.

1. Archaean granitoids 1.1. Diﬀerent types of granites Archaean granitoids belong to one of four broad groups (Champion & Smithies 1999, 2007a; Moyen et al. 2003b). These groups are (1) either muscovite or (less commonly) cordierite bearing peraluminous S-type granites; (2) alkaline to peralkaline granites and associated syenites; (3) magnesiopotassic granitoids, including sanukitoids (Shirey & Hanson 1984; Stern et al. 1989; Stern & Hanson 1991; Smithies & Champion 2000; Moyen et al. 2003b; Martin & Moyen 2005, 2007; Martin et al. 2005); and (4) plagioclase-rich granitoids, or tonalite–trondhjemite–granodiorite–monzogranite (TTGM) group (Feng & Kerrich 1992). In the ﬁrst three groups, alkali feldspar is at least as abundant as plagioclase – these are granites s.s. (i.e. monzogranites and syenogranites) and granodiorites. By contrast, K-feldspar is a subordinate mineral in the last group (plagioclase-rich granitoids), which includes the tonalite–trondhjemite–granodiorite (TTG) series, as well as a large number of ‘post-tectonic potassic granites’ (typically, granodiorites to monzogranites). These plagioclase-rich granitoids are by far the dominant component of Archaean plutonic rocks, and indeed of the preserved Archaean crust in general, perhaps corresponding to as much as 80% or 90% of the preserved Archaean continents. Plagioclase-rich granitoids deﬁne a continuum, between true TTG to more K-rich granites with K2O/Na2O up to ~1·5. It is common practice to separate the TTG (sodic granitoids) from the potassic granites (e.g. Windley 1995); indeed in many Archaean provinces, the TTG mostly occur as part of the ‘background’ orthogneisses, whereas the potassic terms of the TTGM group form late, well-identiﬁed plutons, and some proposed Archaean granite classiﬁcations emphasise this difference (TTG vs. biotite-bearing granites in Moyen et al. 2003b). However, this two-fold classiﬁcation is an oversimpliﬁcation. Closer examination of some ‘late potassic granites’, such as the GMS (granodorite–monzogranite–syenite) suite of the Barberton granite–greenstone terrane, reveals that the supposedly potassic batholiths actually contain a signiﬁcant proportion of sodic (tonalitic and trondhjemitic) rocks, and that they deﬁne a continuum, from these sodic compositions to potassic granites and syenites (Anhaeusser & Robb 1983b; Robb 1983; Westraat et al. 2004; Belcher & Kisters 2006). Symmetrically, many of the ‘grey gneiss complexes’, supposedly dominated by the sodic TTGs, do actually contain a signiﬁcant potassic component, e.g. the granitoid ‘domes’ of the Pilbara terrain, the gneiss complex in the Eastern Dharwar Craton in India, etc. (Moyen et al. 2003a; Champion & Smithies 2007b). Consequently, other attempts at classifying Archean granites have rather emphasised the similarities between rocks in the plagioclase-rich granitoids group, e.g. by referring to the potassic members as LILE-enriched TTGs (Champion & Smithies 2007b). Furthermore, from an experimental perspective, all rocks of the plagioclase-rich granitoids group (and, in fact, all granitoids without a clear mantle connection) probably share a common origin, by melting of a source made of feldspars and hydrous maﬁc silicates (Clemens & Vielzeuf 1987; Vielzeuf & Montel 1994; Patin˜o-Douce & Beard 1995; Montel & Vielzeuf 1997; Stevens et al. 1997; Vielzeuf & Schmidt 2001; Moyen & Stevens 2006). Typical melting reactions correspond to incongruent breakdown of the hydrous minerals: amphibole in the more maﬁc sources (yielding sodic melts), biotite in the more

felsic, enriched sources (giving potassic melts). At higher pressures (>20–25 kbar), where amphibole becomes unstable (Schmidt & Poli 1998), epidote breakdown can play a role, although melting reactions in this part of the P–T space are poorly understood (Skjerlie & Patin˜o-Douce 2002; Patin˜oDouce 2005). For instance, the amphibole breakdown reaction (Rapp et al. 1991; Rapp & Watson 1995; Moyen & Stevens 2006) has the general form plagioclase+amphibole=melt+ clinopyroxene+(orthopyroxene or garnet), where garnet is typically stable at pressures higher than 15 kbar (Fig. 1). This melting behaviour is broadly similar to biotite breakdown reactions (biotite+plagioclase+quartz=melt+(orthopyroxene+ garnet). This reaction (or group of reactions) are relevant to melting over a large range of sources compositions – from relatively primitive tholeiitic amphibolites (amphibole+calcic plagioclase), to more evolved rocks such as tonalites (An30–40 plagioclase, quartz, amphibole and occasional biotite) or even trondhjemites (quartz, An30, biotite). In fact, many aspects of the geochemical discussion that follows are also applicable to the melting of greywacke sources (plagioclase+quartz+biotite) that generate Phanerozoic S-type granites (Clemens & Vielzeuf 1987; Vielzeuf & Montel 1994; Stevens et al. 1997; Vielzeuf & Schmidt 2001). Therefore, all of the Archaean plagioclase-rich granitoids, either sodic or potassic (and actually, all the crustally-derived granitoids, including the S-type granites), form through similar petrogenetic processes, by comparable melting reactions and, importantly, with the same sort of residual minerals; this implies that the resulting melts will have the same geochemical behaviour, and that the same interpretation applies to all of them – from the sodic TTG proper, to the more potassic members of this group.

1.2. Melting reactions and composition of the residuum Whilst dominated by plagioclase, the crustal sources for plagioclase-rich granitoids (and for most granitoids in general) include a hydrous maﬁc mineral. As a result, melting is controlled by very similar reactions tied to the breakdown of either amphibole or biotite. This releases (i) water, which triggers the melting of whatever feldspar is available, and of quartz if present; and (ii) an anhydrous silicate, such as cordierite, garnet or orthopyroxene, or various combinations of these minerals (Stevens et al. 1997). The sodic component of plagioclase preferentially enters the melt (at lower temperatures than the calcic component), so any residual plagioclase will be more calcic than the original feldspar (Rapp et al. 1991; Vielzeuf & Montel 1994; Patin˜o-Douce & Beard 1995; Rapp & Watson 1995; Stevens et al. 1997; Patin˜o-Douce 2005; Moyen & Stevens 2006). As a result, granitoid melts typically coexist with a solid residuum that comprises relatively calcic plagioclase and maﬁc minerals such as cordierite, garnet or pyroxene. At temperatures greater than 1000–1100(C, these minerals are also incorporated into the melt and disappear, although such temperatures are probably uncommon in the crust (Fig. 1). The nature of the residual assemblage is also strongly inﬂuenced by the pressure (depth) of melting. Garnet is stable at pressures greater than 6 to 15 kbar, depending on the composition of the source assemblage, with rocks containing excess aluminium producing garnet much more easily, and at lower depths, compared to Al-poor compositions (Stevens et al. 1997; Moyen et al. 2006). Plagioclase also disappears from the residuum at high pressure; the pattern is more complicated, as in the absence of melt, the plagioclase-out line is positively sloped in the P–T space and its position depends on the anorthite content of the plagioclase. Therefore, sodic plagioclase is intrinsically stable at higher pressures, but is also

GRANITE CHEMISTRY AS A MARKER OF SOURCE AND DEPTH

37

Figure 1 Simpliﬁed phase relations applicable to (a) melting of plagioclase-amphibolebiotite sources (compiled after Moyen & Stevens (2006), with modiﬁcations based on Skjerlie & Patin˜o-Douce (2002) and Patin˜o-Douce (2005) at high pressure) and (b) fractionation of water-saturated tonalitic liquids (Schmidt & Thompson 1996). Geotherms indicated for reference (grey, dashed-dotted) in both ﬁgures. Mineral abbreviations: (amp) amphibole; (bi) biotite; (cpx) clinopyroxene; (epi) epidote; (grt) garnet; (ilm) ilmenite; (opx) orthopyroxene; (phn) phengite; (pl) plagioclase; (q) quartz; (ru) rutile; (zo) zoisite. In (a), the grey ﬁeld represent the ‘band’ in which the multivariant melting reactions occur; solid lines correspond either to mineral stability boundaries (mineral appear on the side of the line which is labelled) or to upper limit of mineral stabilities (in the case of multivariant melting reactions; the mineral are unstable on the side where the mineral abbreviation is parenthesised). [Amphib] and [BPQ] labels denote lines applicable only to either amphibolite or biotite– plagioclase–quartz sources, respectively. Mineral associations coexisting with melt are indicated in italic, black for amphibolite sources and white for BPQ. In (b), the lines correspond to mineral appearance (mineral stable on the side its name is written); the grey ﬁeld denotes the near-solidus region where most of the crystallisation occurs. The arrows correspond to the three possible crystallisation paths discussed in section 3.2, and the mineral proportions used in each model are indicated in the boxes.

consumed by melting at lower temperatures, compared to more calcic plagioclase. The result is a negatively-sloped plagioclase-out line in the P–T space, the position of which depends on the calcic or sodic nature of the rock (more calcic lithologies tend to have an expanded plagioclase stability ﬁeld: Moyen & Stevens 2006). Investigating the nature of the residuum in terms of phase relationships, however, obscures the fact that the residuum continuously changes with changing P–T conditions. Mapping of the phase proportions in the P–T space (Moyen & Stevens 2006) has revealed that (a) above the garnet-in line, the amount of garnet increases with increasing pressures and with decreasing temperatures and, (b) below the plagioclase-out line, the amount of plagioclase decreases with increasing temperatures and pressures.

2. Controls on the composition of Archaean plagioclase-rich granitoids (TTGM): composition of primary melts 2.1. The chemistry of primary melts For trace elements, the link between the composition of the source and of the melt is expressed in the batch melting equation: Cl ⫽

C0 F ⫹ D.共1 ⫺ F兲

where Cl is the concentration of an element in the liquid and C0 the concentration of that element in the source (Shaw

1970). The melt fraction (F) is largely a function of the conditions of melting – i.e. of temperature, H2O content and, to a lesser degree, pressure. D is the bulk partition coeﬃcient, and is expressed as D⫽

兺K X

i D i

i

where Xi is proportion of mineral i in the residue and KDi is the partition coeﬃcient of an element between melt and that mineral. While the KD values are not strictly constant over the P–T-melt composition space, their variations are suﬃciently minor to be ‘masked’ by the overall uncertainties of the model (see the discussion in Be´dard 2010), especially with a broad approach as used in the present paper, and so constant values are used (Table 1). The mineral proportions in the residuum are a function of the conditions of melting (pressure, temperature and water activity) as well as the nature of the source. Therefore, Shaw’s equation as written above expresses the fact that the composition of a (granitoid) melt is a function of the source’s composition, and of the P–T–H2O conditions of melting. Most crustal rocks will produce very similar residuum assemblages when they melt (plagioclase, orthopyroxene and garnet being the dominant minerals) (Fig. 1) This means that, for the elements with a high KD in the restite-making minerals, the nature of the protolith actually exerts a much smaller eﬀect on the composition of a granitoid melt than the P–T conditions of melting. In most cases, crustal melting will occur under ﬂuid-absent conditions associated with the breakdown of hydrous phases like biotite or amphibole, such that the

38

J.-F. MOYEN ET AL.

variation of the water content is relatively minor. In any case, higher water contents would usually lead to higher melt fractions without substantially altering the nature of the residual phases, and so the eﬀect that higher water contents would have on the composition of a granitoid melt would be indistinguishable from that of higher temperatures. This implies that the trace element composition of primary granitic melts is primarily controlled by only three parameters: 1. The temperature of melting, or more precisely the degree of melting (as higher pressures tend to decrease the melt fraction, whereas higher water contents increase it). High melt fraction liquids are more maﬁc – i.e., have lower concentrations of SiO2 and higher MgO and FeO. Most elements are strongly correlated with SiO2, resulting in narrow, linear trends in Harker diagrams. Accordingly, the eﬀect of SiO2-correlated geochemical variation to a granitoid (or granitic suites) ‘geochemical signal’ should ideally be ﬁltered out before any discussion on the geochemical diﬀerences or similarities between granitic melts. 2. The composition of the source, especially with regard to its concentration of incompatible elements. In the case of a strongly incompatible element (Dz0), the batch melting equation approaches Cl =C0/F: i.e., the concentration of that element in the melt is only a function of the melt fraction and the concentration of that element in the source. In Harker diagrams, melts from diﬀerent source compositions therefore deﬁne stacked parallel trends, with melts from richer sources plotting towards higher compositions. LILE (Rb, Ba, Th), LREE and some of the HFSE (Th, Zr, Nb) behave in this way. Signiﬁcantly, K also behaves as an incompatible trace element in alkali feldspar-free and biotite-free assemblages, such as the source of the plagioclase-rich granites considered here, and its concentration in these melts is likewise aﬀectively only a function of source composition and the degree of melting. However, K-enriched sources will yield liquids in which K behaves as a major element, deﬁning K-rich trends that do lead to alkali feldspar bearing rocks that extend from tonalites to granodiorites and to monzogranites. Therefore, depending on the degree of K-enrichment, ‘plagioclase-rich’ granitoids will deﬁne either ‘tonalite–trondhjemite’ or ‘tonalite–granodiorite’ series. 3. The depth of melting. Recent work focused on the critical examination of experimental data on partial melting of amphibolites (Moyen & Stevens 2006) showed that two minerals, whose stability and abundance is strongly pressure-dependant, play a key role in controlling the concentration of important trace elements during melting of plagioclase-amphibole dominated assemblages. Plagioclase disappears at pressures in excess of 10–20 kbar, depending on temperature and source composition (etc.), and controls the concentration of Sr and Eu (or more interestingly, the magnitude of the Eu anomaly) in the melt. Melts coexisting with plagioclase are relatively Sr- and Eu-poor, owing to the large KD of plagioclase for this element. To some extent, Al (and to a lesser degree Na) will also be preferentially partitioned into plagioclase during melting. Garnet appears at ca. 12 kbar and becomes increasingly abundant with increasing pressure (towards eclogitic assemblages) and controls the concentration of HREE and Y in the melt, such that melts in equilibrium with garnet are Y and HREE poor. The combined eﬀect of these two minerals during melting at progressively higher pressures is an evolution from HREE- and Y-rich melts that are Sr-poor and have a negative Eu anomaly, to relatively HREE- and Y-poor melts with higher Sr contents and no Eu anomaly. The

eﬀect that source enrichment has on the concentration of these trace elements in a melt is secondary compared to the control exerted by the absence or presence of these mineral phases. Again, melts formed at diﬀerent pressures deﬁne stacked trends in Harker diagrams; the degree of melting primarily controls the absolute concentrations and the position within a trend. The expected result is that melts will deﬁne ‘trends’ in Harker-type diagrams (Fig. 2), where the SiO2 and compatible elements contents will be tightly correlated and reﬂect the degree of melting; melts from diﬀerent depths will yield ‘stacked’ trends in diagrams using pressure-sensitive elements such as Sr, Y or Nb (but will be superposed on diagrams using either very incompatible elements, or purely compatible elements); melts from diﬀerent sources will deﬁne stacked trends in diagrams using incompatible elements (but will be superposed when looking at pressure-sensitive or compatible elements).

2.2. A test based on experimental data In order to demonstrate the reliability of this approach, it was tested on granitic melts from known sources generated under known conditions of melting. Such a situation does not occur naturally, but can be produced in controlled experiments. One diﬃculty here is that trace elements are rarely analysed in experimental melts, such that a real, direct comparison is not feasible. Nevertheless, experimental charges are probably the one situation in which the conditions of equilibrium and of closed systems are realised, and therefore perhaps correspond to the only situation where Shaw’s (1970) batch melting equation can reliably be applied to calculate the melt composition. Therefore, data on modal compositions from >350 experiments (compiled by Moyen & Stevens 2006) and published partition coeﬃcients (Table 1) were used to calculate the trace element compositions of the corresponding melts. Three diﬀerent source compositions were used; one corresponds to a normal, to slightly enriched MORB (mid-ocean ridge basalt), such as the one proposed for Archaean MORBs (Jahn et al. 1980; Condie 1981; Jahn 1994); another, more enriched source, has a composition close to that of an arc basalt; whilst the third source corresponds to a tonalitic composition. Obviously, melting of a tonalite would not give exactly the same melting reactions, melt fractions, etc., and using this composition is not strictly rigorous. The point here, however, is merely to test the inﬂuence of contrasting sources compositions. This simple model conﬁrms the qualitative conclusions outlined above. Two main classes of elements are evident. Elements such as Rb, Th, etc. are primarily controlled by the nature of the source, and pressure exerts little or no inﬂuence on their concentration within the melt. On the other hand, for Sr, Y and Nb, pressure appears as the dominant parameter, and the source-related variations, while playing a role, remain second-order. Clearly, the two parameters (pressure and enrichment) are not completely independent (Fig. 3). For example, at pressures where plagioclase is not stable, Sr-enriched sources will yield Sr-richer melts compared to Sr-depleted sources. In contrast, at low pressures, plagioclase is stable and high Sr contents in a melt are impossible to achieve regardless of the Sr concentration of the source. On the other hand, while Y is clearly pressure-controlled, high Y concentrations can be achieved only at low pressures (no garnet in the residuum); but low Y concentrations can exist both at high pressure (given a garnetbearing residuum) or at low pressures (for a Y poor source). Diagrams employing elements that are controlled by source enrichment alone (or nearly so – e.g. Rb, Th or K2O/

GRANITE CHEMISTRY AS A MARKER OF SOURCE AND DEPTH

39

Figure 2 Expected positions of trends generated by partial melting in Harker type diagrams. (a) Diagrams using compatible elements (Mg, Ni, Cr) will yield tightly correlated trends that reﬂect the melt amount, and do not give information regarding the depth of melting or the nature of the source. Fractionation will move the composition to the right, whereas melting and solids entrainment (section 3.1) will displace the composition to the left, keeping it in the same, narrow trend; (b) Diagrams using incompatible elements (Th, Rb, K) show stacked trends, which position reﬂects the source enrichment; (c) Diagrams using pressure-sensitive elements (Sr, Nb, Y) produce stacked trends, and their position gives information on the depth of melting. Table 1 Partition coeﬃcients and source concentrations used in the model. Partition coeﬃcients from (Foley et al. 2000; Foley et al. 2002; Schmidt et al. 2004; Be´dard 2006), using values applicable for intermediate to felsic liquids in equilibrium with amphiboleplagioclasegarnet. Partition coeﬃcients

Ba Rb Th Nb Ta La Ce Sr Nd P Sm Zr Hf Y Yb

Source compositions (ppm)

Amphibole

Clinopyroxene

Garnet

Ilmenite

Orthopyroxene

Plagioclase

Rutile

Olivine

MORB-like

Arc basalt

Tonalitic

0·046 0·055 0·055 0·274 0·477 0·319 0·56 0·389 1·32 0·225 2·09 0·417 0·781 2·47 1·79

0·006 0·01 0·104 0·007 0·028 0·028 0·059 0·032 0·115 0·162 0·259 0·125 0·208 0·603 0·635

0·0004 0·0007 0·0075 0·04 0·08 0·028 0·08 0·019 0·222 0·184 1·43 0·537 0·431 14·1 23·2

0·018 0·025 0·09 3 2·7 0·015 0·012 0·0022 0·01 0·002 0·009 2·3 2·4 0·037 0·13

0·047 0·047 0·13 0·01 0·126 0·0003 0·0007 0·047 0·0028 0·05 0·0085 0·031 0·246 0·054 0·125

1·016 0·068 0·095 0·239 0·053 0·358 0·339 6·65 0·289 0·079 0·237 0·078 0·069 0·138 0·094

0·0043 0·0076 0·2 n.a. n.a. 0·0057 0·0065 0·036 0·0082 0·03 0·0954 3·7 4·97 0·0118 0·0126

0·0205 0·0231 0·0542 0·0103 0·1260 0·0216 0·0185 0·0306 0·0123 0·0357 0·0091 0·0365 0·0195 0·0301 0·0581

10 1 0·015 3·5 0·2 3 8 100 8 520 3·5 80 2·4 28 3·3

50 10 0·5 3·507 0·2 5 12 300 11·2 570 3·75 104·24 2·974 35·82 3·9

690 55 6·9 10 0·5 32 56 454 21·4 — 3·3 152 3·9 7·5 0·55

Na2O) – can yield unambiguous information (Fig. 2). Likewise, diagrams using elements whose concentration is strongly pressure-dependant (Sr, Y, HREE) give slightly more ambiguous, but still useful, results. However, most other elements do not exhibit such speciﬁc behaviour. (Either there are no phases

with high enough KD values to allow for spectacular pressurerelated changes, or these elements are not incompatible enough to be used as good source tracers.) Diagrams using these elements are typically harder to interpret, but are, in any case, of no relevance in the present paper since the aim is to establish

40

J.-F. MOYEN ET AL.

Figure 3 Pressure vs. composition diagrams for experimental melts. All panels show pressure vs. element diagrams, recalculated from experimental data. Black symbols correspond to pressures >15 kbar, open symbols <15 kbar; triangles are tonalitic sources, circles are ‘arc-like’ and squares ‘MORB-like’ (Table 1). The left three panels show typical source-controlled elements, whereas the right hand side panels show typical depth-controlled elements, with the key mineral indicated on the panel. Trace elements in ppm and majors in wt.%, in this ﬁgure and all the following.

a tool to derive information on the pressure of melting and the enrichment of the source. Another way to show the pressure eﬀects on melt compositions is the use of multi-element diagrams (spidergrams, Fig. 4). With increasing pressures of melting, the depth of the Nb–Ta anomaly increases, the negative Sr anomaly is progres-

sively reduced, and the Y–Yb segment on the right-hand side is progressively depressed. These eﬀects respectively correspond to the increasing role of rutile, the decreasing plagioclase contents, and the increasing proportion of residual garnet, and in this case are directly correlated to changes in the abundance of these minerals. In contrast, variations in source composition

GRANITE CHEMISTRY AS A MARKER OF SOURCE AND DEPTH

41

Figure 4 Multi-elements diagrams (Thompson 1982) for calculated experimental melts in diﬀerent pressure bands. The source used is the MORB-like composition (Table 2). Panels (a)–(d) show diﬀerent pressure bands (panel (e)). in each case the grey background ﬁeld indicate the whole spread of data. Panel (e) summarises which mineral controls diﬀerent sections of the spectrum: rutile for the Nb–Ta, through plagioclase for the Sr anomaly and garnet for the Tb–Yb section.

result in stacked but parallel patterns in spidergrams, especially in the left-hand part (more incompatible elements) of the diagrams. In Harker diagrams (Fig. 5), the behaviour described here is also evident, and stacked trends are produced as predicted. Each pair (source composition, depth band) yields a speciﬁc ‘trend’. Diagrams using a ‘depth-controlled’ element produce stacked trends, separated according to pressure (diﬀerent source compositions giving superposed trends). In contrast, diagrams using an ‘enrichment-controlled’ element give trends separated by the degree of enrichment. As stated above, the separation is actually much better in the case of the source control, than in the depth control. In either case, moving along the trend gives information on the degree of melting (F), with more maﬁc compositions corresponding to higher degree of melting.

3. From primary melts to granites The previous discussion focused solely on the composition of primary granitoid melts. It is likely, however, that granitoids (the rocks themselves) do not represent pure liquid composition, and certainly do not reﬂect primary melt compositions. Several processes can combine to modify the composition of primary melts. These include entrainment of solid phases (restite as well as entrained crystals that formed from the melt itself), mixing with other (mantle-derived?) magmas, assimilation of country rock, and fractional crystallisation. Therefore, it may be diﬃcult to directly link the granite’s composition with source conditions.

3.1. Entrainment of solids In granitoids in general, and in the Archaean crustal (plagioclase-rich) granitoids studied in the present paper,

solids entrained from the source with the magma can aﬀect the bulk rock geochemistry of granitoids (eﬀectively, solid-liquid mixtures). This can be either easy to identify in the ﬁeld, when the granitoids studied are actually diatexites (Sawyer 1998; Moyen et al. 2003a; Be´dard 2006); or more cryptic, through processes such as restite unmixing (Chappell et al. 1987) or entrainment of peritectic minerals (Stevens et al. 2007). In all cases however, the net, geochemical eﬀect of these processes is to add part of the source to the melt; the resulting compositions are therefore intermediate between the pure melt and the source. In other words, the net result is to extend the geochemical trend towards the source (the apparent temperature or F, as observed from the trends, will be higher); on the other hand, distinct trends will remain clearly separated (Fig. 2).

3.2. Fractional crystallisation Fractional crystallisation has the potential to aﬀect the composition of any magmatic rock, and its inﬂuence must be assessed. Indeed, on pure geochemical (trace elements) grounds it is possible to dramatically increase e.g. the Sr/Y and La/Yb ratios of melts, therefore moving magmas from one trend to another one (e.g. Be´dard 2006). However, it seems extremely unlikely that fractionation plays a signiﬁcant role in shaping the geochemical features of TTGM – and in particular, in explaining the diﬀerences between diﬀerent trends: (1) In granites in general, there is now a growing consensus that fractionation plays only a limited role (Clemens et al. 2010), and indeed there is no well-documented example for which geochemical trends in granitoids can be ascribed to fractional crystallisation. One does not see why Archaean granitoids would be an exception to this; (2) When well identiﬁed, mappable TTG plutons are studied (e.g. the 3·45 Ga Stolzburg pluton of Barberton (Moyen et al. 2007)),

42

J.-F. MOYEN ET AL.

Figure 5 Harker diagrams for one pressure-dependant (Th) and one source-dependant (Sr) element. Compositions recalculated from experimental data. On the left (same caption as for Fig. 3), colours indicate the depth and symbols the nature of the source. Stacked trends as described in the text are obvious, reﬂecting clearly the diﬀerent sources for Th, less clearly the depth of melting for Sr (note that there are actually two low-Th trends, that appear superposed due to the low overall abundances, but are nevertheless one order of magnitude diﬀerent, with typical values <0·1 for the depleted MORB source and ca. 1 ppm for the arc source). On the right, the same diagrams, plus SiO2 vs. FeO+MgO, are drawn using symbols corresponding to diﬀerent temperatures; the temperature is broadly correlated to the position in the trend (scatter is due to diﬀerences related to the water content, the depth of melting, the nature of the source, etc.)

GRANITE CHEMISTRY AS A MARKER OF SOURCE AND DEPTH

43

Figure 6 Eﬀects of fractional crystallisation. The low pressure (LP) and high pressure (HP) domains correspond to experimental compositions, compiled from the literature or recalculated as described in the text (section 2.2); HP is >15 kbar and LP<15 kbar, ‘isotherms’ are schematically depicted, from generalisation of the data from Figure 5. The arrows correspond to the three fractionation models (‘FC model’) discussed in the text (section 3.2; see also Fig. 1).

the range of compositions – that could possibly be ascribed to fractionation – is actually restricted. Likewise, whenever fractional crystallisation was studied and modelled for major and traces together (Martin 1987), it turned out that it produced only small, short trends (a few SiO2 percent points, from 67% to 72%). This probably puts an empirical upper limit on the eﬀect of fractionation; (3) While fractional crystallisation can indeed change signiﬁcantly the trace elements compositions of the melts (especially if phases such as amphibole or garnet do fractionate), the degree of fractionation (especially of maﬁc minerals) is strongly limited by the available Fe and Mg; even the most primitive TTGs rarely exceed 6–7% of FeO+MgO, meaning that no more than 30% of a typical, 15–25% FeO+MgO, cumulate (Moyen et al. 2007) can fractionate – this does strongly limit the potential of fractionation; (4) Finally, in order to strongly aﬀect the Sr, Y or Yb contents, the cumulate would need to be garnet-rich and therefore to correspond to high pressure fractionation: in other words, an high pressure evolution is required to arrive to compositions similar to the high Sr series, be it from high pressure melting or high pressure fractionation. To illustrate the last two points above, models of fractionation were calculated at diﬀerent pressures. Based on phase diagram for water-saturated tonalitic melts and experimental results (Schmidt & Thompson 1996; Fig. 1), three possible cooling and fractionation paths were identiﬁed: (1) a high pressure (>15 kbar) evolution, in which clinopyroxene, the liquidus phase, is replaced by epidote and amphibole through peritectic reactions; this results in the fractionation of epidote, garnet and amphibole in proportions of approximately 1:2:4. (2) A medium pressure (ca. 10 kbar) evolution, where a similar peritectic reaction occurs, below the garnet stability ﬁeld; epidote and amphibole crystallise in proportions close to 1:5. (3) A low pressure (<10 kbar) evolution, corresponding to crystallisation during emplacement within the crust, where the fractionating phases are amphibole and plagioclase in approximately equal proportions, with traces of ilmenite; this is similar to the case modelled by Martin (1987). In each case, the amount of crystallisation that can be achieved before running out of Fe and Mg is actually small, in the range of 20–30%. This is consistent with the geochemical models of Martin (1987) and does also match the experimental

results of Schmidt and Thompson (1996): most of the crystallisation occurs close to the solidus, in the ﬁeld where quartz and feldspars form. The resulting vectors are plotted on Figure 6, together with the ﬁelds obtained for experimental melts as described above. The longest possible trends are the one produced by low-pressure (syn-emplacement?) evolution and fractionation of amphibole + plagioclase; even these are actually short, compared to the range that can be generated by melting trends. Furthermore, the fractionation vectors are largely superposed to the melting trends. This actually underlines the fact that melting and fractionation are eﬀectively symmetrical processes, and thermodynamically reversible reactions: for given P–T conditions, the melting reaction, say Amp+Pl= Melt+Opx at low pressure, is eﬀectively the same as the fractionation reaction (that consumes the early formed Opx in peritectic reactions) (Stevens et al. 1997). In theory both processes are diﬀerent – crystallisation is generally regarded as ‘fractional’ and therefore as a disequilibrium process, with instantaneous removal of crystals formed, whereas melting is supposed to be mostly an equilibrium process. Geochemical common wisdom therefore leads to treating both processes in diﬀerent ways (and with diﬀerent equations). However, given the increasing recognition that granitic melts have viscosities too high to permit signiﬁcant solid/melt separation (Giordano et al. 2008), and the ﬁeld and petrological evidence that granites are poorly separated mushes (solid+crystals) (Weinberg 2006; Stevens et al. 2007), one may seriously question the ‘fractional’ nature of crystallisation in granites. If both melting and fractionation are mostly equilibrium processes (or processes with the same ‘amount of disequilibrium’, so to say), they are therefore perfectly symmetrical and undistinguishable on geochemical grounds. The trends observed for a given unit (one magma batch) can therefore be a combination of melting, or crystallisation, or both. As long as both occur at the same or similar depths, the melting and crystallisation trends will be superposed. It is therefore concluded (1) that the role of fractionation is actually a second order eﬀect, compared to the role of melting; and that the trends deﬁned by Archaean TTGM are, at least in ﬁrst approximation, essentially melting trends; and (2) that even when fractionation plays a role, it will aﬀect the ‘within

44

J.-F. MOYEN ET AL.

trend’ variability, but not the ‘inter-trend’ diversity. Once generated, granitic melts remain conﬁned to their trend, regardless of subsequent modiﬁcations.

3.3. Mixing with mantle-derived melts Mixing with melts from diﬀerent origins (typically, mantle derived melts, such as sanukitoids) has the greatest potential to move the compositions away from the main, melting-related trends. Therefore, all the interpretation proposed in the present paper (and in particular, the diagrams discussed below (see section 4)) holds if, and only if, it can be demonstrated that the melts are purely crustal melts, with no mantle component. In other words, the interpretation proposed must not be used indistinctly, but rather at the end of a comprehensive petrological study. Some indications that mantle melts (or interactions with the mantle) did play a role, and therefore that the present analyses do not hold, are: (1) ﬁeld evidences such as microgranular maﬁc enclaves (Didier et al. 1982; Didier & Barbarin 1991); and (2) geochemical evidence, as particularly well expressed in sanukitoids (Shirey & Hanson 1984; Stern et al. 1989; Stern & Hanson 1991; Smithies & Champion 2000; Moyen et al. 2003b; Martin et al. 2005, 2010; Martin & Moyen 2005, 2007). Coupled enrichment in LILE and compatible elements, i.e. high K2O values for low SiO2 and high MgO (as an empirical criteria) K2O contents >2%, for MgO>2%, is a reasonably strong indication of a mantle component (Miller et al. 2008), high contents in Rb, Th, Ni, Cr and Sr at the same time, etc.

3.4. The meaning of geochemical trends The discussion above shows that some care must be exercised in using the geochemistry of granitic rocks to discuss source processes, and that the approach outlined here should not replace a careful petrological and geochemical interpretation. However, if the composition of individual samples from a suite can vary considerably, the ‘trend’ (be it a melting, or a diﬀerentiation, controlled trend) that the suite as a whole forms provides the best reﬂection of the nature of the source and of the melting conditions. The shapes and positions of the trends (here, in Harker type diagrams) are primarily controlled by the nature of the source and the depth of melting, provided some assumptions (e.g., no mantle inﬂuence, etc.) can be demonstrated to be true. The diﬀerences between granites of diﬀerent origins is typically larger than the diﬀerences induced by post-melting processes: i.e. the petrogenetic processes that form the primary melt typically have a compositional signature that survives the petrogenetic processes that subsequently modify primary melt compositions. On the other hand, the very nature of the ‘trend’ approach erases any information on the degree of melting (and/or of fractionation, and/or of solid entrainment); these processes are not resolved by this approach, as they essentially result in moving compositions along the trends, but not from one to another.

4. Interpreting the geochemical signal in terms of depth and enrichment As demonstrated, granitoid series from diﬀerent origins (in terms of source or pressure of melting) should deﬁne ‘stacked’ trends in Harker diagrams. Diagrams using incompatible elements (LILE, K2O, LREE, HFSE such as Th, Zr or Nb) will reﬂect the degree of source enrichment, whereas diagrams using Sr, Na, Al or Eu will give an indication of plagioclase stability or instability, and diagrams using HREE or Y will

Figure 7 Building delta-diagrams. Rocks from diﬀerent granite suites will typically depict stacked trends in Harker-type diagrams (X vs. SiO2), such that the absolute value of the concentration in an element X is ambiguous. On the other hand, the trends are distinctive, and reﬂect diﬀerent degrees of enrichment or depth of melting. By calculating, for each sample, the distance between the analysis and a reference line, the delta parameters mostly reﬂects the trend to which a sample belong – more than its absolute concentration.

reﬂect the absence or presence of garnet – the latter two being proxies for the pressure of melting. Devising diagrams showing both characteristics at the same time is problematic. For example, diagrams that simply plot one ‘source-enrichment tracer’ against one ‘depth tracer’ (e.g., plotting La vs. Yb or Rb vs. Sr) will be complicated by the fact that each series evolves along its own trend (in Harker diagrams), and that all elements vary as a function of SiO2 – resulting in large overlaps for granites from diﬀerent origins. Trace element ratios are commonly used in geochemistry to avoid this problem, and this is why diagrams such as Sr/Y vs. Y or La/Yb vs. Yb largely behave as pressure indicators. In the present paper a new set of diagrams is developed that can simultaneously reveal information relating both to source enrichment (composition) and to pressure of melting. These diagrams explicitly remove the contribution of SiO2-related evolution, and use as an indicator a new parameter X, where X is any given element. For each element, X is taken as the distance between the analysed value, and a reference line, in a SiO2–X (Harker) diagram (Fig. 7). The choice of the reference line is not critical, but because these diagrams were developed initially to assist the interpretation of Archaean plagioclaserich granites, the divide between the Pilbara high- and low-Sr series (Champion & Smithies 2007b) was chosen as a reference (Fig. 8). Using another line with the same slope would just uniformly move the values up or down, without changing their relative position; using another line with a diﬀerent slope will have greater eﬀects. On the other hand, as long as the reference line chosen is one that ﬁts the global evolution and shape of the geochemical trends, the diﬀerences will be relatively small. Problems would arise only if the reference line was at a large angle with the trend’s slope; in this case, the delta parameter would vary with SiO2, therefore negating the expected gains. Plotting SiO2 against delta (not shown) does help in testing whether the reference line used is correct (the resulting diagram should yield ﬂat or nearly ﬂat trends), and the parameters of the reference line were tested and adjusted by trial and error.

GRANITE CHEMISTRY AS A MARKER OF SOURCE AND DEPTH

45

Figure 8 Harker diagrams for Pilbara and Barberton rocks. The dashed grey line is the reference line used for calculating the delta parameter. Stacked trends, reﬂecting either diﬀerent sources (Rb, Th, K2O/Na2O) or the depth of melting (Sr, Y, Al2O3, Nb to some degree) are evident. Pilbara data (Champion & Smithies 2007b) are in grey and Barberton data (Moyen et al. 2007) in black. Diﬀerent symbols correspond to diﬀerent plutons as in Figure 10. Table 2 Constants used in the calculation of delta parameters. For each element, X=X(a SiO2 +b). a and b are empirically estimated, by using a reference line (plotted in Fig. 8) that separates Pilbara’s diﬀerent sub-series (Champion et al. 2007b).

Sr Y Nb Th Rb A2O3

a

b

20 1·25 0·35 0·5 5 0·25

1700 100 33 20 220 32

If the equation of the reference line is written as X=a SiO2 +b, we deﬁne X=X(a SiO2 +b) (Table 2). In theory, for very incompatible elements (D=0), the evolution line related to melting (or fractionation) should rather be a hyperbolic or exponential curve (as their concentration is Cl =C0/F,

with F being a function of SiO2), and it would be more rigorous to use such a curve as a reference line. In practice, however, this does not result in a signiﬁcant improvement over ‘linear’ ﬁts. It was also found that K could be readily substituted by the simpler K2O/Na2O parameter, since Na concentrations are not strongly aﬀected by the degree of source enrichment; they are aﬀected by the depth of melting, but this eﬀect is small compared to the eﬀect source enrichment has on the K2O/Na2O ratio. Most of the variance in the geochemical signal is related to diﬀerentiation, with SiO2 contributing most of it. The diagrams presented aim at eliminating the variance related to the diﬀerentiation component, to focus on the smaller components of the variance. In essence, this approach is fairly similar to the ‘sliding normalisation’ of Lie´geois et al. (1998), or the ‘oxide*’ parameters of Bonin (1986). Diagrams plotting A vs. B – where A is a depthcontrolled element (Sr, Y, HREE) and B is an enrichmentcontrolled element (LILE, LREE, some HFSE) – can therefore be interpreted as a ‘depth of melting vs. source enrichment

46

J.-F. MOYEN ET AL.

Figure 9 Identiﬁcation of granitoid groups with diﬀerent geochemical signatures, using three units as an example. The three examples used are the ca. 3·45 Ga plutons in Barberton (Moyen et al. 2007); the ca. 3·45 Ga Mount Edgar suite of the Pilbara craton (Champion & Smithies 2007b), and the ca. 3·3 Ga Carbaa suite of the Pilbara. The grey ﬁeld corresponds to the ﬁeld of post-Archaean S-type granites (compilation from the literature, >350 analyses). Diagrams use either the delta parameter as explained in the text, or K2O/Na2O. On each diagram, the arrows show qualitatively the eﬀects of pressure and source enrichment. In all diagrams, there is a clear split between two angled trends, one going towards deep sources (relatively depleted) and one towards shallower, but richer, sources. S-type granites typically plot along the shallow/rich trend.

diagram’. As discussed above, eﬀects relating to the melt fraction, or indeed to further diﬀerentiation, essentially result in moving the compositions along one trend and therefore have been mostly removed from the geochemical signal. However, such diagrams should not be used indiscriminately and do not replace a comprehensive petrological analyses; one must be especially aware of potential problems such as (1) the role of mantle components, that will falsify many of the assumptions these diagrams are based on; or (2) a-typical sources or processes, resulting in an ambiguous (or misleading) geochemical signal. A telltale sign would be the de-correlation between geochemical indicators supposedly carrying the same information (i.e., Sr and Y, or Rb and K); such decorrelations should always be treated as a sign of potential ‘problems’ and evidence that some of the basic assumptions of the method do not hold.

5. Application to meso-Archaean granites from Barberton and the Pilbara Delta diagrams were applied to granites from the two reasonably well-known meso-Archaean (3·5–3·1 Ga) provinces, the Barberton granite–greenstone terrane of South Africa (BGGT) (Moyen et al. 2007) and the Eastern Pilbara Craton of Australia (Champion & Smithies 2007b). These provinces are dominated by seemingly similar granitoids.

5.1. Diﬀerent types of TTGM associations Using the delta diagrams devised as explained above provides a convenient way to compare the geochemistry of diﬀerent granitoids; compared to Harker-type diagrams, the interpretation is easier, as most of the diﬀerentiation component (i.e., SiO2-related scatter) is removed; compared to classical spidergrams, the resulting diagrams are less cluttered. Delta diagrams using a range of enrichment-related (K2O/Na2O, Rb, Th) and pressure-related (Sr, Nb, Y, Al2O3) elements were ﬁrst used to illustrate the fact that diﬀerent series have very distinctive geochemical signatures. Figure 9 shows the delta diagram for three plutonic units in Barberton and the Pilbara craton; they were chosen because they show fairly extreme geochemical characteristics, and are eﬀectively representative end-members that illustrate the validity of this approach. For comparison, the ﬁeld for Phanerozoic S-type granites was also plotted on these diagrams. These rocks derive from potassic metasediment-dominated sources, where K-feldspar and possibly muscovite or aluminium silicates play a signiﬁcant role during melting. Predictably, they plot in the ‘low-pressure, enriched sources’ region of the diagrams, consistent with their origin by intracrustal melting. The ca. 3·45 Ga-old plutons from Barberton (Theespruit, Stolzburg) are characterised by their position in the ‘deep, depleted’ quarter of the diagrams; this reﬂects deep melting of

GRANITE CHEMISTRY AS A MARKER OF SOURCE AND DEPTH

a fairly depleted (or not enriched) source (Clemens et al. 2006; Moyen et al. 2007). The ca. 3·3 Ga Carbana suite in the Pilbara Craton (Champion & Smithies 2007b) shows an opposite behaviour (shallow melting of enriched sources), in fact largely overlapping with the ﬁeld of S-type granites. Between the two, the 3·45 Ga Mont Edgar granitoids (Pilbara: Champion & Smithies 2007b) reﬂect the shallower melting of a source maybe marginally more enriched than the source of Barberton examples.

5.2. Regional histories of granitoid plutonism Delta diagrams provide a simple tool to discuss and compare the geochemistry of plutonic suites, and more importantly, to understand their geodynamical signiﬁcation. As an illustration, the evolution of granitoid compositions between ca. 3·5 Ga and 3·2 Ga is now compared, in both Barberton and the Pilbara. In Barberton, three successive, well-deﬁned plutonic cycles are known before 3·2 Ga (Anhaeusser & Robb 1980, 1983a; Anhaeusser et al. 1983; Clemens et al. 2006; Moyen et al. 2007): + At 3·55–3·51 Ga, a small composite pluton (the Steynsdorp pluton) made of tonalites and granodiorites is formed south-east of the main greenstone belt; + At 3·45 Ga, the trondhjemitic plutons of the ‘Stolzburg domain’ are formed (mainly the Stolzburg and Theespruit pluton, but also some smaller occurrences). + At 3·27–3·21 Ga, large trondhjemitic to tonalitic plutons are formed west and north of the greenstone belt (the Badplaas, Nelshoogte and Kaap Valley plutons). A more complicated pattern emerges for the Eastern Pilbara Craton. There are four successive ‘supersuites’ older than 3·1 Ga, but these do not show systematic relative geographic distribution – i.e. rocks from all suites occur throughout the terrain (Champion & Smithies 2007b). These suites are the Callina supersuite (3·45–3·5 Ga), the Tambina supersuite (3·45–3·4 Ga, broadly synchronous with the Stolzburg and associated plutons in Barberton), the Emu Pool supersuite (ca. 3·3 Ga) and the Cleland supersuite (3·3–3·2 Ga, synchronous with the Badplaas, Nelshoogte and Kaap Valley plutons). Within each of the supersuites, both ‘high Al’ and ‘low Al’ types have been identiﬁed and both ‘LILE-enriched’ and ‘normal’ types are present. Champion & Smithies (2007b) and Moyen et al. (2007) provide a detailed petrogenetic interpretation of the Pilbara and Barberton granites. Harker diagrams for all the mesoArchaean rocks (Fig. 8) show clear ‘stacked’ trends. Very high Sr concentrations distinguish the Barberton 3·45 Ga group, whilst the low-Al, low-Sr 3·3 Ga Emu Pool suite of the Pilbara Craton is also clearly identiﬁed. Likewise, the high K/Na, and corresponding high Rb, of part of Pilbara’s Emu Pool suite, and of Barberton’s GMS, is clearly illustrated. ‘Delta’ diagrams illustrate very clear diﬀerences between the supersuites (Fig. 10), but allow these diﬀerences to be more readily interpreted in terms of petrogenesis. Collectively, the data deﬁne ‘angled’ trends separating two end-member processes: deep melting of depleted sources (melting along a subducted slab?) and shallow melting of richer sources (intracrustal melting?). Rocks with compositions that can be interpreted as reﬂecting either deep melting of rich sources or shallow melting of depleted sources appear to be less common. Three main types of granitoids are represented in the two provinces, as described previously:

47

+ The majority of the samples belong to a group evolving by shallow or medium-pressure melting of depleted to poorly enriched sources. The 3·45 Ga Mt Edgar and Shaw suites in the Pilbara belong to this type, as do most of the suites from the 3·3 Ga Emu Pool supersuite, and in Barberton, most of the Steynsdorp pluton and of the ca. 3·2 Ga intrusive rocks. + The second type, commonly associated, is represented by shallow melts of enriched sources. They form a minority component in the Steynsdorp and 3·2 Ga plutons of Barberton (Badplaas in particular); and are abundant in the Emu Pool (3·3 Ga) and Cleland (3·25 Ga) supersuites of the Pilbara. Champion & Smithies (2007b) proposed that this ‘high-LILE granite group’ from the Pilbara Craton suite are a result of recycling an enriched, felsic crustal source, likely over a range of pressures lower than those required to stabilise abundant garnet and destabilise plagioclase. + The ca. 3·45 Ga granite group from Barberton is distinctive in typically having a combination of higher Sr and lower Y and Yb than other granites. This was interpreted to reﬂect high-pressure melting of a depleted (metabasaltic) source – the most commonly invoked model for Archean granites of variably TTG-like composition. The crustal recycling component (melting of an enriched, felsic source) that is proposed for some of the Pilbara rocks (high-LILE), and the Barberton GMS suite, is well reﬂected by high K2O/Na2O, high Rb values, corresponding to horizontal extensions along the X-axes in the diagrams. It also correlates with low-pressure signatures. A high-pressure origin for the ca. 3·45 Ga group in Barberton is also clearly established, as is the depleted nature of their source, as they plot along the Y-axes. As outlined above, the ‘depth-related’ signal tends to be less clear than the ‘source related’ signal in delta diagrams. This is obvious in diagrams using Sr or Y as a depth-related parameter, in which the ‘enrichment’ trend is slightly oblique to, rather than parallel with, the X-axes (in other words, there is a positive correlation between Rb or K/Na, and Sr or Y). In geodynamic terms, it is noticeable that the granitoids geochemistry gives a fairly diﬀerent picture for both cratons, even though they form in a broadly synchronous way (Zegers et al. 1998). The Steynsdorp event (3·55–3·51 Ga) has no documented equivalent in the Pilbara. The ca. 3·45 Ga event is represented in Barberton by a short-lived burst of highpressure granitoids (Stolzburg, Theespruit), but corresponds in the Pilbara to a much longer-lived (80 Ma, Callina and Tambina Supersuites) period of melts generated from possibly similar sources, but at shallower depths. The massive ca. 3·3 Ga event (the Emu Pool supersuite in the Pilbara), that is proposed to be concomitant with the main some-forming event (Van Kranendonk et al. 2007), consists in mostly shallow melting of mixed sources; it has no match in Barberton, except the small (and poorly documented) Stentor Pluton (Kamo & Davis 1994). Finally, both cratons do show some plutonic activity at 3·25–3·2 Ga; whereas it is the major event in Barberton, where a range of compositions reﬂect time and space shifts in melting regions (Moyen et al. 2007), it is only a minor component (the Cleland Supersuite) of the evolution of the East Pilbara, in which only enriched shallow sources are involved.

6. Conclusion The ﬁrst order source of variance in the composition of Archaean crustal (plagioclase-rich, or TTGM) granitoids is related to the position of individual samples on ‘diﬀerentiation trends’ – regardless of the origin of such trends. However, this

48

J.-F. MOYEN ET AL.

Figure 10 Delta diagrams for Barberton and Pilbara rocks. Due to space constraint, only one diagram (K2O/Na2O vs. Sr) was plotted for each suite; this diagram is chosen as being one of the most informative (compare Fig. 9), and also because these three elements are analysed in all the datasets, even for samples from the literature whose composition was obtained before the advent of cheap trace elements analyses. Barberton samples on the left, Pilbara on the right; samples are grouped by chronological order, bottom to top. Discussion in text.

GRANITE CHEMISTRY AS A MARKER OF SOURCE AND DEPTH

range of composition carries little information on petrogenetic processes that are ‘hidden’ in second-order variations. The relevant petrogenetic information (source enrichment and depth of melting) is ‘hidden’ behind the diﬀerentiation history, and is reﬂected by the position of the trend to which a sample belongs, rather than the composition of that individual sample. Constructing diagrams that allow the geochemistry of granitic rocks to be described as belonging to a speciﬁc trend, rather than in terms of absolute composition, directly facilitates interpretations about source composition as well the depth of melting. This in turn places constrains on the geometry, and to some degree the P–T condition, of a crustal segment during granite formation. Applying that approach to the mid-Archaean Barberton and Pilbara granites reveals unexpected petrogenetic diversity for magmas typically referred to as ‘TTGs’. Melting depths as well as source compositions vary widely, such that generalisations about the compositions and origins of ‘TTGs’ become rather meaningless. These revelations, when considered in conjunction with the contrasting way particular compositional groups of granites are spatially and temporally distributed within each of these contemporaneous cratons, should lead to more robust interpretations regarding the distinct geodynamic environments that may have prevailed.

7. Acknowledgements JFM’s work in Barberton was funded through a NRF grant awarded to A. Kisters, University of Stellenbosch (grant no. NRF 2053186), and a ‘starting grant’ by the University of Stellenbosch.

8. References Anhaeusser, C. R. & Robb, L. J. 1980. Regional and detailed ﬁeld and geochemical studies of archean trondhjemitic gneisses, migmatites and greenstone xenoliths in the southern part of the Barberton mountain land, South Africa. Precambrian Research 11, 373–97. Anhaeusser, C. R., Robb, L. J. & Viljoen, M. J. 1983. Notes on the Provisional geological map of the Barberton greenstone belt and surrounding granitic terrane, eastern Transvaal and Swaziland (1:250 000 colour map). In Anhaeusser, C. R. (ed.) Contributions to the geology of the Barberton Mountain Land. National Geodynamic Programme, Barberton Project. Geological Society of South Africa Special Publication 9, 221–3. Anhaeusser, C. R. & Robb, L. J. 1983a. Chemical analyses of granitoid rocks from the Barbeton Mountain Land. In Anhaeusser, C. R. (ed.) Contributions to the geology of the Barberton Mountain Land. National Geodynamic Programme, Barberton Project. Geological Society of South Africa Special Publication 9, 189–219. Anhaeusser, C. R. & Robb, L. J. 1983b. Geological and geochemical characteristics of the Heeenveen and Mpuluzi batholiths south of the Barberton greenstone belt, and preliminary thoughts on their petrogenesis. In Anhaeusser, C. R. (ed.) Contributions to the geology of the Barberton Mountain Land. National Geodynamic Programme, Barberton Project. Geological Society of South Africa Special Publication 9, 131–51. Be´dard, J. 2006. A catalytic delamination-driven model for coupled genesis of Archaean crust and sub-continental lithospheric mantle. Geochimica et Cosmochimica Acta 70, 1188–214. Be´dard, J. 2010. Parental magmas of Grenville Province massif-type anorthosites, and conjectures about why massif anorthosites are restricted to the Proterozoic. Earth and Environmental Science Transactions of the Royal Society of Edinburgh 100 (for 2009), 77–103. Belcher, R. W. & Kisters, A. F. M. 2006. Progressive adjustments of ascent and emplacement controls during incremental construction of the 3·1 Ga Heerenveen batholith, South Africa. Journal of Structural Geology 28, 1406–21. Bonin, B. 1986. Ring complexes and anorogenic magmatism. Amsterdam: Elsevier. Champion, D. C. & Smithies, R. H. 1999. Archaean granites of the Yilgarn and Pilbara cratons: secular changes. In Barbarin, B. (ed.)

49

The Origin of Granites and Related Rocks, Fourth Hutton Symposium Abstracts, 137. Clermont-Ferrand, France: BRGM. Champion, D. C. & Smithies, R. H. 2007a. Three billion years of granite magmatism: Paleoarchaean to Permian granites of Australia. In Miller, J. A. & Kisters, A. F. M. (eds) 6th International Hutton Symposium Abstract Volume & Program Guide, 63–64. Stellenbosch, South Africa: Department of Geology, Geography & Environmental Sciences, University of Stellenbosch. 236 pp. Champion, D. C. & Smithies, R. H. 2007b. Geochemistry of Paleoarchean granites of the East Pilbara terrane, Pilbara Craton, Western Australia: implications for early Archean crustal growth. In Van Kranendonk, M. J., Smithies, R. H. & Bennet, V. (eds) Earth’s Oldest Rocks. Developments in Precambrian Geology 15, 369–410. Amsterdam: Elsevier. Chappell, B. W., White, A. J. R. & Wyborn, D. 1987. The Importance of Residual Source Material (Restite) in Granite Petrogenesis. Journal of Petrology 28 (6), 1111–38. Clemens, J. D., Yearron, L. M. & Stevens, G. 2006. Barberton (South Africa) TTG magmas: geochemical and experimental constraints on source-rock petrology, pressure of formation and tectonic setting. Precambrian Research 151, 53–78. Clemens, J. D., Helps, P. A. & Stevens, G. 2010. Chemical structure in granitic magmas – a signal from the source? Earth and Environmental Science Transactions of the Royal Society of Edinburgh 100 (for 2009), 159–72. Clemens, J. C. & Vielzeuf, D. 1987. Constraints on melting and magma production in the crust. Earth and Planetary Science Letters 86, 287–306. Collins, W. J. 2007. Using the mantle to unravel granite geodynamics. In Miller, J. A. & Kisters, A. F. M. (eds) 6th International Hutton Symposium Abstract Volume & Program Guide, 72–73. Stellenbosch, South Africa: Department of Geology, Geography & Environmental Sciences, University of Stellenbosch, 236 pp. Condie, K. C. 1981. Archean greenstone belts. Amsterdam: Elsevier. Didier, J. & Barbarin, B. 1991. Enclaves and Granite Petrology. Amsterdam: Elsevier. Didier, J., Duthou, J. L. & Lameyre, J. 1982. Mantle and crustal granites: genetic classiﬁcation of orogenic granites and the nature of their enclaves. Journal of Volcanology and Geothermal Research 14, 125–32. Feng, R. & Kerrich, R. 1992. Geochemical evolution of granitoids from the Archean Abitibi southern volcanic zone and the Pontiac subprovince, Superior Province, Canada: implications for tectonic history and source regions. Chemical Geology 98, 23–70. Foley, S. F., Barth, M. G. & Jenner, G. A. 2000. Rutile/melt partition coeﬃcients for trace elements and an assessment of the inﬂuence of rutile on the trace element characteristics of subduction zone magmas. Geochimica et Cosmochimica Acta 64, 933–8. Foley, S. F., Tiepolo, M. & Vannucci, R. 2002. Growth of early continental crust controlled by melting of amphibolite in subduction zones. Nature 417, 637–40. Giordano, G., Russel, J. & Dingwell, D. B. 2008. Viscosity of magmatic liquids: a model. Earth and Planetary Science Letters 271, 123–34. Jahn, B. 1994. Ge´ochimie des granitoı¨des arche´ens et de la crouˆte primitive. In Hagemann, R., Jouzel, J., Treuil, M. & Turpin, L. (eds) La ge´ochimie de la Terre. CEA-Masson. Jahn, B., Auvray, B., Blais, S., Capdevila, R., Cornichet, J., Vidal, F. & Hammeurt, J. 1980. Trace elements geochemistry and petrogenesis of Finnish greenstone belts. Journal of Petrology 21, 201–44. Kamo, S. L. & Davis, D. W. 1994. Reassessment of Archean Crustal Development in the Barberton Mountain Land, South-Africa, Based on U–Pb Dating. Tectonics 13 (1), 167–92. Lie´geois, J. P., Navez, J., Hertogen, J. & Black, R. 1998. Contrasting origin of post-collisional high-K calc-alkaline and shoshonitic versus alkaline and peralkaline granitoids. The use of sliding normalization. Lithos 45(1–4), 1–28. Martin, H. 1987. Petrogenesis of Archaean trondhjemites, tonalites and granodiorites from eastern Finland; major and trace element geochemistry. Journal of Petrology 28 (5), 921–53. Martin, H. 1994. The Archean grey gneisses and the genesis of the continental crust. In Condie, K. C. (ed.) Archean crustal evolution. Developments in Precambrian Geology 11, 205–59. Amsterdam: Elsevier. Martin, H., Smithies, R. H., Rapp, R. P., Moyen, J.-F. & Champion, D. C. 2005. An overview of adakite, tonalite–trondhjemite– granodiorite (TTG) and sanukitoid: relationships and some implications for crustal evolution. Lithos 79 (1–2), 1–24.

50

J.-F. MOYEN ET AL.

Martin, H., Moyen, J.-F. & Rapp, R. 2010. The sanukitoids series: magmatism at the Archaean–Proterozoic transition. Earth and Environmental Science Transactions of the Royal Society of Edinburgh 100 (for 2009), 15–33. Martin, H. & Moyen, J.-F. 2005. The Archaean–Proterozoic transition: sanukitoid and Closepet-type magmatism. Mineralogocial Society of Poland Special Papers 26, 57–68. Martin, H. & Moyen, J.-F. 2007. Sanukitoid and Closepet-type magmatism: the Archaean–Proterozoic transition. In Miller, J. A. & Kisters, A. F. M. (eds) 6th International Hutton Symposium Abstract Volume & Program Guide, 129–30. Stellenbosch, South Africa: Department of Geology, Geography & Environmental Sicences, University of Stellenbosch. 236 pp. Miller, J. A., Moyen, J.-F. & Benn, K. 2008. The 2·74–2·66 Ga Kenogamissi complex (Abitibi): evolving sources of plutons mirroring geodynamics. Geochimica et Cosmochimica Acta 72 (12S), A629. Montel, J. M. & Vielzeuf, D. 1997. Partial melting of metagreywackes, part II. Compositions of minerals and melts. Contributions to Mineralogy and Petrology 128, 176–96. Moyen, J.-F., Martin, H., Jayananda, M. & Peucat, J.-J. 2003a. Magmatism during the accretion of the late Archaean Dharwar Craton (South India): sanukitoids and related rocks in their geological context. Abstracts from EGS–AGU–EUG Joint Assembly, Nice, France, 6–11 April 2003, Abstract EAE03-A-00516. Moyen, J.-F., Martin, H., Jayananda, M. & Auvray, B. 2003b. Late Archaean granites: a typology based on the Dharwar Craton (India). Precambrian Research 127 (1–3), 103–23. Moyen, J.-F., Stevens, G. & Kisters, A. F. M. 2006. 3·2 Ga highpressure, low-temperature metamorphism in the Barberton greenstone belt: the evidence for Archaean mountain belts and subduction zones. In Condie, K. C., Kro¨ner, A. & Stein, R. J. (eds) When did plate tectonics begin on Earth? Theoretical and empirical constraints. GSA Penrose Conference, Lander, Wyoming, 13–18 June 2006. Boulder, Colorado: Geological Society of America. Moyen, J.-F., Stevens, G., Kisters, A. F. M. & Belcher, R. W. 2007. TTG plutons of the Barberton granitoid–greenstone terrain, South Africa. In Van Kranendonk, M. J., Smithies, R. H. & Bennet, V. (eds) Earth’s Oldest rocks. Developments in Precambrian geology, 606–68. Amsterdam: Elsevier. Moyen, J.-F. & Stevens, G. 2006. Experimental constraints on TTG petrogenesis: implications for Archean geodynamics. In Benn, K., Mareschal, J.-C. & Condie, K. C. (eds) Archean geodynamics and environments. AGU Geophysical Monograph 164, 149–78. Washington, DC: American Geophysical Union. Patin˜o-Douce, A. E. & Beard, J. S. 1995. Dehydration-melting of Bt gneiss and Qtz amphibolite from 3 to 15 kB. Journal of Petrology 36, 707–38. Patin˜o-Douce, A. E. 2005. Vapor-absent melting of tonalite at 15– 32 kbar. Journal of Petrology 46 (2), 275–90. Rapp, R. P., Watson, E. B. & Miller, C. F. 1991. Partial melting of amphibolite/eclogite and the origin of Archaean trondhjemites and tonalites. Precambrian Research 51, 1–25. Rapp, R. P. & Watson, E. B. 1995. Dehydration melting of metabasalt at 8–32 kbar: implications for continental growth and crustmantle recycling. Journal of Petrology 36 (4), 891–931. Robb, L. J. 1983. Geological and geochemical characteristics of late granite plutons in the Barberton region and Swaziland, with an emphasis on the Dalmein pluton – a review. In Anhaeusser, C. R. (ed.) Contributions to the geology of the Barberton Mountain Land. National Geodynamic Programme, Barberton Project. Geological Society of South Africa Special Publication 9, 153–67. Sawyer, E. W. 1998. Formation and evolution of granite magmas during crustal reworking; the signiﬁcance of diatexites. Journal of Petrology 39 (6), 1147–67.

Schmidt, M. W., Dardon, A., Chazot, G. & Vannucci, R. 2004. The dependence of Nb and Ta rutile-melt partitioning on melt composition and Nb/Ta fractionation during subduction processes. Earth and Planetary Science Letters 226, 415–32. Schmidt, M. W. & Poli, S. 1998. Experimentally based water budgets for dehydrating slabs and consequences for arc magma generation. Earth and Planetary Science Letters 163, 361–79. Schmidt, M. W. & Thompson, A. B. 1996. Epidote in calc-alkaline magmas: an experimental study of stability, phase relationships and the role of epidote in magmatic evolution. American Mineralogist 81, 462–74. Shaw, D. M. 1970. Trace Element Fractionation During Anatexis. Geochimica et Cosmochimica Acta 34 (2), 237–43. Shirey, S. B. & Hanson, G. N. 1984. Mantle-derived Archaean monzodiorites and trachyandesites. Nature 310, 222–4. Skjerlie, K. & Patin˜o-Douce, A. E. 2002. The ﬂuid-absent partial melting of a zoisite bearing quartz eclogite from 1·0 to 3·2 GPA: implications for melting of a thickened continental crust and for subduction-zone processes. Journal of Petrology 43, 291–314. Smithies, R. H. & Champion, D. C. 2000. The Archaean high-Mg diorite suite: Links to Tonalite–Trondhjemite–Granodiorite magmatism and implications for early Archaean crustal growth. Journal of Petrology 41 (12), 1653–71. Stern, R. A., Hanson, G. N. & Shirey, S. B. 1989. Petrogenesis of mantle-derived, LILE- enriched Archean monzodiorites and trachyandesites (sanukitoids) in Southwestern Superior Province. Canadian Journal of Earth Sciences 26, 1688–712. Stern, R. A. & Hanson, G. N. 1991. Archaean high-Mg granodiorites: a derivative of light rare earth enriched monzodiorite of mantle origin. Journal of Petrology 32, 201–38. Stevens, G., Clemens, J. D. & Droop, G. T. R. 1997. Melt production during granulite-facies anatexis: experimental data from ‘primitive’ metasedimentary protoliths. Contributions to Mineralogy and Petrology 128, 352–70. Stevens, G., Villaros, A. & Moyen, J.-F. 2007. Selective peritectic garnet entrainment as the origin of geochemical diversity in S-type granites. Geology 35 (1), 9–12. Thompson, R. N. 1982. British tertiary volcanic province. Scottish Journal of Geology 18, 49–107. Van Kranendonk, M. J., Hickman, A. H., Smithies, R. H. & Champion, D. C. 2007. Paleoarchean development of a continental nucleus: the East Pilbara terrane of the Pilbara craton, Western Australia. In Van Kranendonk, M. J., Smithies, R. H. & Bennet, V. (eds) Earth’s Oldest rocks. Developments in Precambrian Geology 15, 307–37. Amsterdam: Elsevier. Vielzeuf, D. & Montel, J. M. 1994. Partial melting of metagreyackes. Part I. Fluid-absent experiments and phase relationships. Contributions to Mineralogy and Petrology 117, 375–93. Vielzeuf, D. & Schmidt, M. W. 2001. Melting reactions in hydrous systems revisited: application to metapelites, metagreywackes and metabasalts. Contributions to Mineralogy and Petrology 141, 251–67. Weinberg, R. 2006. Melt segregation structures in granitic plutons. Geology 34, 305–8. Westraat, J. D., Kisters, A. F. M., Poujol, M. & Stevens, G. 2004. Transcurrent shearing, granite sheeting and the incremental construction of the tabular 3·1 Ga Mpuluzi batholith, Barberton granite–greenstone terrane, South Africa. Journal of the Geological Society, London 161, 1–16. Windley, B. F. 1995. The Evolving Continents. Chichester: John Wiley & Sons. Zegers, T. E., de Wit, M. J., Dann, J. & White, S. H. 1998. Vaalbara, Earth’s oldest assembled continent? A combined structural, geochronological, and palaeomagnetic test. Terra Nova 10 (5), 250–9.

MS received 12 March 2007. Accepted for publication 23 September 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 51–60, 2010 (for 2009)

Similarities between mantle-derived A-type granites and voluminous rhyolites in continental ﬂood basalt provinces Simon Turner and Tracy Rushmer 1

GEMOC, Department of Earth and Planetary Sciences, Macquarie University, Sydney NSW 2109, Australia Email: [email protected]; [email protected]

ABSTRACT: Many continental ﬂood basalt provinces contain rhyolites with ‘A-type’ compositions and many studies have concluded that these higher silica rocks are crustal melts from metapelitic or tonalitic country rock. However, although many of the low-Ti continental ﬂood basalt sequences exhibit a marked a silica gap from w55–65 wt.% SiO2, many incompatible element ratios, and the calculated eruption temperatures (950–1100(C) are strikingly similar between the rhyolites and associated basalts. Using experimental evidence, derivation of the low-Ti rhyolites from a basaltic parent is shown to be a viable alternative to local crustal melting. Comparison of liquid compositions from experimental melting of both crustal and mantle-derived (basaltic) source materials allows the two to be distinguished on the basis of Al2O3 and FeO content. The basalt experiments are reversible, such that the same melts can be produced by melting or crystallisation. The eﬀect of increased water content in the source is also detectable in the liquid composition. The majority of rhyolites from continental ﬂood basalt provinces fall along the experimental trend for basalt melting/ crystallisation at relatively low water content. The onset of the silica gap in the rhyolites is accompanied by an abrupt decrease in TiO2 and FeO*, marking the start of Fe–Ti oxide crystallisation. Diﬀerentiation from 55–65 wt.% SiO2 requires w30% fractional crystallisation in which magnetite is an important phase, sometimes accompanied by limited crustal contamination. The rapid increase in silica occurs over a small temperature interval and for relatively small changes in the amount of fractional crystallisation, thus intermediate compositions are less likely to be sampled. It is argued that the presence of a silica gap is not diagnostic of a crustal melting origin for either A-type granites or rhyolites in continental ﬂood basalt provinces. The volume of these rhyolites erupted over the Phanerozoic is signiﬁcant and models for crustal growth should take this substantial contribution from the mantle into account. KEY WORDS: rhyolite

A-type granite, continental ﬂood basalts, fractionation, melting experiments,

Whether silicic magmatic rocks are largely derived by partial melting or crystal fractionation is a much discussed question and one closely linked to debates over the origin of the continental crust. One of the striking observations is that most silica-rich magmatic rocks often occur in bimodal associations. In orogenic settings these include distinctive suites termed A-type rocks (e.g. Collins et al. 1982; Eby 1990). Most continental ﬂood basalt provinces contain, and frequently are capped by, extensive rhyolite sequences (see Mahoney & Coﬃn 1997 for a recent review). The presence of a silica gap in these bimodal suites has been widely regarded as evidence against a liquid line of descent in bimodal magmatic suites, with the lack of intermediate rocks put down to the fact that crustal rocks are broadly granitic in composition and therefore tend to yield rhyolitic melts (e.g. Leeman 1982; Bellieni et al. 1986; Sylvester 1989). Instead the higher silica rocks have frequently been interpreted as crustal melts, with the heat for melt generation being derived from the emplacement of the associated basaltic magmas (e.g. Huppert & Sparks 1988). In cases where the higher silica rocks have isotope ratios similar to those of the basalts, it has been argued that they were derived by remelting of the intrusive equivalents of the basalts (e.g. in the Lebombo, Cleverly et al. 1984; the Deccan, Lightfoot et al. 1987; and the Parana´, Piccirillo et al. 1987). In the present paper, it is shown

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016181

that these rhyolites share many important characteristics with A-type granites and volcanic rocks, and geochemical and experimental data are combined to argue that fractional crystallisation from the basalt itself, not crustal melting of associated metapelite or intermediate rock types, is the most likely mechanism of generation of many of the high-temperature, high-silica magmatic suites.

1. Comparison of A-type granites with Rhyolites from continental flood basalt provinces Table 1 shows average major element compositions of A-type granites and silicic rocks from continental ﬂood basalt provinces worldwide, along with some estimates of eruption temperatures and volumes. A-type granites and their volcanic equivalents typically occur in bimodal magmatic suites, and at least some can be shown to be the result of fractionation of accompanying basaltic magma (Turner et al. 1992; Frost & Frost 1997; McCurry et al. 2008; Whitaker et al. 2008). In addition to elevated concentrations of SiO2 and incompatible elements combined with low Al2O3 and CaO (see Table 1), A-type rocks are also distinguished from the more common Iand S-type igneous rocks by their high magmatic temperatures and relatively anhydrous mineral assemblages (e.g. Clemens

– 940 1,2

– –

73·60 0·33 12·69 2·90 0·06 0·33 1·09 3·34 4·51 0·09

1200 1000 3

890000 20000

68·85 0·89 12·72 5·95 0·10 1·05 2·47 2·88 4·23 0·25

low-Ti Parana´

1200 980 4

50000 small

66·69 0·93 11·38 4·72 0·14 0·35 3·72 2·65 3·08 0·25

Tasmania granophyre

– – 5,6

250000 15000

69·81 0·76 12·97 4·07 nd 0·75 2·01 3·11 4·43 0·26

British Tertiary Province

– – 7,8

unknown unknown

71·60 0·39 12·28 3·55 nd 0·43 0·90 2·97 5·46 0·06

Keweenawan

– – 9

175000 10000

74·40 0·90 13·00 1·60 nd 0·30 0·50 2·60 6·60 0·40

Columbia River (glass)

1200 1100 10

1000000 35000

63·05 0·76 16·06 6·31 0·16 2·54 3·20 2·70 2·04 0·19

Karoo (central)

1200 1000 11,12

14 1

73·80 0·31 13·07 1·64 0·05 0·01 0·51 3·52 6·20 0·02

Black Hill granophyre

1200 1000 3

89000 20000

65·90 1·29 13·25 7·00 0·13 1·27 2·73 3·42 4·25 0·40

high-T Parana´

– – 10

1000000 35000

69·39 0·50 12·43 6·26 0·11 0·33 1·46 3·14 4·53 0·14

Lebombo rhyolite

1050 900 13

500000 500

65·63 0·66 15·39 5·12 nd 1·24 1·47 4·51 5·05 0·11

Deccan Salsette island

– – 14

unknown unknown

71·02 0·55 12·54 5·19 nd 0·56 0·67 2·57 5·11 0·29

Huronian

– 1060 15

unknown 3000

67·09 0·74 14·24 4·08 0·10 0·84 2·47 3·43 4·75 0·20

Yardea dacite

1: Collins et al. (1982); 2: Clemens et al. (1986); 3: Garland et al. (1995); 4: Hergt et al. (1987); 5: Walsh et al. (1979); 6: Gamble et al. (1992); 7: Green & Fitz (1993); 8: Nicholson & Shirey (1990); 9: Lambert et al. (1989); 10: Milner & Duncan (pers. comm. 1996); 11: Turner (1996); 12: Turner & Foden (1996); 13: Lightfoot et al. (1987); 14: Jolly et al. (1992); 15: Creaser & White (1991).

Volume (km3) basalt rhyolite Temperature ((C) basalt rhylote Reference(s)

SiO2 (wt.%) TiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K2 O P2O5

A-type granite average

Table 1 Typical compositions of A-type granites, CFB rhyolites (including volumes and average temperatures of rhyolites and associated basalts) and granophyre.

52 SIMON TURNER AND TRACY RUSHMER

A-TYPE GRANITES AND RHYOLITES IN FLOOD BASALTS

53

(Fig. 1). It is these many similarities which suggest the likelihood of a common petrogenesis which we now explore through comparison with experimental data. Initial comparisons are made with the well documented low-Ti rhyolites from the Parana´-Etendeka province. These have incompatible trace element and isotopic characteristics which are nearly identical to underlying low-Ti basalts and which both Garland et al. (1995) and Ewart et al. (2004) have used to argue for a fractional crystallisation origin for the rhyolites.

2. Experimental investigation of crustal melting

Figure 1 Rb vs. Y+Nb granite discrimination diagram from Pearce et al. (1984) showing the compositional overlap between A-type granites (data from Collins et al. 1982; Turner et al. 1992; Turner & Foden 1996) and rhyolites from continental ﬂood basalt provinces (data sources as in Table 1). BTVP=British Tertiary volcanic province; VAG=volcanic arc granites; ORG=orogenic granites; syn-COLG= syn-collision granites; WPG=within-plate granites.

et al. 1986; Turner et al. 1992). On trace element discrimination diagrams, A-type granites largely fall in the within-plate granite ﬁeld of Pearce et al. (1984) (Fig. 1). Volumes can be non-trivial and in Australia, for example, the Gawler Range Volcanics are a suite of high temperature A-type volcanic rocks with an estimated volume of >3000 km3 (e.g. Creaser & White 1991). Bimodal basalt-rhyolite suites are common in continental ﬂood basalt provinces, generally with high-temperature rhyolite units capping the basalts (the Parana´-Etendeka, the Karoo, the Deccan Trap and the Proterozoic of the Keweenawan and Huronian), but can also contain intrusive silicic equivalents at lower temperatures (the British Tertiary province, and the Ferrar province of Antarctica and Tasmania). Characteristic features include large aerial extent, generally anhydrous primary mineral assemblages, and often a high proportion of glass in the matrix. The mode of eruption of this distinctive class of rhyolites is not easy to determine, as high silica magmas tend to be highly viscous and unable to ﬂow long distances. Possible causes of the lowered viscosity are either an abundance of volatiles (Holtz & Johannes 1994), or unusually high eruption temperatures or eﬀusion rates (Henry et al. 1988). Alternatively the rhyolite units could have been emplaced as pyroclastic eruptions, or as hybrid ‘rheoignimbrites’ in which the initially pyroclastic material is rewelded on contact with the ground due to the sustained high temperatures (Fig. 2). Where large calderas are found with the hightemperature rhyolites, a pyroclastic origin is favoured (e.g. Milner & Ewart 1989), but in a number of cases there is no such evidence (e.g. Ekren et al. 1984). Rhyolites associated with continental ﬂood basalt provinces tend to be devoid of traces of eruptive centres, thus their mode of eruption is not easy to explain. Whilst it is obviously diﬃcult to generalise about the chemistry of high-temperature rhyolites world-wide, there are a number of major element features which characterise this group. In particular, the high-temperature rhyolites tend to have moderate silica contents, low Al2O3, TiO2, MgO and FeO* that are often almost identical to those of A-type granites (see Table 1). They also have elevated incompatible trace element concentrations and overlap with A-type granites

Numerous experimental studies have been performed which should allow the distinction of partial melts of diﬀerent crustal lithologies. The likely source materials of crustal melts are metasediments and quartzo-felspathic, intermediate rock types (e.g. tonalite) in the mid–upper crust, and granulites, basalts or amphibolites in the lower crust. Experimental work on partial melting of such rocks with varying source composition, pressure, water content and temperature has been carried out in a number of studies, and the major element composition of the melt determined (e.g. Helz 1976; Spulber & Rutherford 1983; Rutter & Wyllie 1988; Vielzeuf & Holloway 1988; Beard & Lofgren 1989, 1991; Rapp et al. 1991; Rapp & Watson 1995; Rushmer 1991; Patin˜o-Douce & Johnston 1991; Skjerlie & Johnston 1993, Sisson et al. 2005). The liquid composition is controlled by the melting reactions in the source, which are in turn determined by factors such as water content, pressure and fO2. Water content of the source arguably plays the most fundamental role in determining the speciﬁc melting reactions, and hence the liquid composition (Beard & Lofgren 1991). Three main scenarios exist: (i) water-saturated melting, where there is free water present; (ii) dehydration or ﬂuid-absent melting, in which water is contained in hydrous phases such as micas and amphiboles, but is not of a suﬃcient quantity to saturate the melt; and (iii) dry melting in which no hydrous phases are involved. The presence of water lowers the granite solidus (Holtz & Johannes 1994), therefore the water content in the source region determines the temperature at which the ﬁrst melt will be produced. Water content also aﬀects the percentage of melt obtained, for example during ﬂuid-absent melting, an assemblage containing micas is more water-rich than one containing an equivalent proportion of amphibole as the only hydrous phase, and this results in 50% as opposed to 20% melt under the same pressure-temperature conditions (Clemens & Vielzeuf 1987). Alternatively, increases in pressure serve to reduce the melt proportion, because at higher pressures, the solubility of water is increased in the melt phase and this reduces melt volume. The eﬀects of fO2 are allied to the water content, as water increases the fO2 by introducing more oxygen into the system (Morse 1980). Figure 3 compiles FeO vs. Al2O3 experimental data from amphibolite melting experiments (Beard & Lofgren 1989, 1991; Spulber & Rutherford 1983; Helz 1976; Rapp et al. 1991; Rushmer 1991; Sisson et al. 2005), tonalites (Rutter & Wyllie 1988; Skjerlie & Johnston 1993) and metasediments (Patin˜oDouce & Johnston 1991; Vielzeuf & Holloway 1988). Within the amphibolite melting region, most data show general trends from low FeO and Al2O3 to higher FeO and Al2O3 as a function of increasing temperature and melt fraction (with the exception of Rapp et al. 1991, where the trends are not as clear). The present authors compare data where fO2 is approximately QFM (Sisson et al. 2005 show the eﬀects of changing fO2). This is also true for Spulber & Rutherford 1983, where experiments are performed at PH2O%Ptotal (region labelled 1); for the PH2O =1 kb Ptotal of Beard & Lofgren (1989; region

54

SIMON TURNER AND TRACY RUSHMER

Figure 2 Photomicrograph of a rhyolite from part of the Parana´ continental ﬂood basalt province in Uruguay showing rheoignimbritic textures (note quartz phenocryst in ﬂattened pumice fragment). Scale bar=1 mm, plane polarised light.

2.1. Metasedimentary and quartzo-felspathic source materials

Figure 3 Al2O3 vs. FeO plot for experimental data on basalts (Helz 1976; Spulber & Rutherford 1983; Beard & Lofgren 1989, 1991; Rapp et al. 1991; Rushmer 1991), tonalites (Rutter & Wyllie 1988; Skjerlie & Johnston 1993) and metasediments (Vielzeuf & Holloway 1988; Patin˜o-Douce & Johnston 1991). Increasing water content in the numbered experimental source basalts: (1) PH2O%Ptotal (Spulber & Rutherford 1983); (2) PH2O =Ptotal (Beard & Lofgren 1989); (3) PH2O =5 kb (Heltz 1976). The Parana´ rhyolites plot at low Al2O3, and moderately high FeO, coinciding with the least wet basalt experiments. Alternatively low pressure plagioclase crystallisation and removal allows the Parana´ rhyolite composition to be reached either by 10% plagioclase crystallisation from the basalts with intermediate water content (2), or by 30% plagioclase crystallisation from the metapelite. See text for further discussion.

labelled 2) and for PH2O =5 kb of Heltz (1976; region labelled 3). Tonalite partial melts have lower FeO and Al2O3 and are comparable to the metapelite melting experiments of Patin˜oDouce & Johnston (1991). Vielzeuf & Holloway’s (1988) metapelite partial melts are distinctly higher in FeO and Al2O3. Both show an increase in FeO and Al2O3 as a function of increasing temperature (see below for further discussion). The Parana´ rhyolites are also plotted, as discussed below.

The eﬀect of higher water content in the source is known to increase the Al2O3, and decrease the FeO abundances of the melt (e.g. Thy et al. 1990). This is due to the decreased stability of plagioclase under wet conditions, while pyroxene stability is correspondingly increased (Beard & Lofgren 1991). Metasedimentary sources tend to contain a substantial proportion of micas (both muscovite and biotite) which are rich in water compared to quartzo-felspathic sources such as tonalite (Clemens & Vielzeuf 1987; Rutter & Wyllie 1988); therefore melts derived from metapelites have higher Al2O3 contents than melts derived from tonalites (Fig. 3). A series of experiments carried out on a metapelite source deﬁcient in plagioclase (Patin˜o-Douce & Johnston 1991) resulted in low Al2O3 abundance (white ﬁeld in Fig. 3), despite a high proportion of mica in the source (40%). However, this can be explained by the production of garnet during the biotite ﬂuid-absent melting reaction, which withholds some Al2O3 from the melt. Despite the variation in Al2O3 content, the metapelites all tend to yield melts with low FeO abundances (generally <3 wt.% FeO), similar to the abundance of FeO in wet melts of basalt sources (Helz 1976). The exception is the melts produced at high temperatures (1050–1250(C) from the metapelitic source as shown by Vielzeuf & Holloway (1988), which started with a high FeO content. This appears to be due to the breakdown of garnet to spinel and quartz, liberating FeO into the melt (in addition to Al2O3, giving a relatively high abundance of both elements). In comparison with the Parana´ rhyolites, melts from metapelites and tonalites are high in Al2O3 and low in FeO, making them unsuitable analogues for the rhyolites. However, the rhyolite magma could also have undergone fractional crystallisation in the crust prior to eruption, in which case the original melt composition would have been more primitive than the samples analysed. The main mineral phase likely to fractionate

A-TYPE GRANITES AND RHYOLITES IN FLOOD BASALTS

55

from silicic magmas over the temperature range 950–1200(C is plagioclase (Skjerlie & Johnston 1993), which would result in a decrease in abundance of Al2O3, CaO and Na2O. On that basis, the high temperature melts from the metapelites of Vielzeuf & Holloway (1988) would appear to be the most suitable parental magma for the rhyolites in terms of Al2O3 and FeO. The Parana´ rhyolites might then reﬂect 30% plagioclase crystallisation from such melts (Fig. 3). However, the CaO and Na2O abundances of the metapelite melts are already low compared to the rhyolites, and are therefore inconsistent with subsequent plagioclase removal. The low FeO content of the metapelite melts of Patin˜o-Douce & Johnston (1991) and the tonalite melts of Skjerlie & Johnston (1993) at similar Al2O3 contents to the rhyolite magmas would require accumulation of pyroxene or Fe–Ti oxides in order to achieve similarity to the rhyolites, which is not considered likely. Thus, none of these sedimentary/quartzo-feldspathic melts satisfy the chemical requirements of the Parana´ rhyolites, and they are considered to be unsuitable source materials for the rhyolite magmas.

2.2. Basalt source An alternative source for the Parana´ rhyolites is derivation from basaltic material, or its hydrated and metamorphosed equivalent (amphibolite). The geographic association of low-Ti rhyolites with low-Ti basalts and high-Ti rhyolites with high-Ti basalts strongly suggests a link between the maﬁc and felsic lavas. Experiments on basalts have been carried out under varying degrees of hydration, with a range in Al2O3 content as mentioned above. The hydrous experiments of Helz (1976) were undertaken at a water pressure of 5 kb, giving rise to melts very rich in Al2O3 (generally R19 wt.% Al2O3) due to the lowered stability of plagioclase (Fig. 3, region 3). Beard & Lofgren (1989) performed experiments on a range of predominantly tholeiitic amphibolites in which PH2O =Ptotal (1 kb) (Fig. 3, region 2). Spulber & Rutherford (1983) used tholeiitic basalts, with experiments carried out over a range of pressures, in which PH2O%Ptotal, making them less water-rich than the experiments of Beard & Lofgren (1991) at pressures greater than 1 kb (Fig. 3, region 1). The high Al2O3 content of the water-saturated melts discounts them as being parental to the Parana´ rhyolites, whilst the melts from the experiments of Spulber & Rutherford (1983) and Beard & Lofgren (1989, 1991) are both potential candidates (regions 1 and 2 respectively in Figure 3). In both cases, the basaltic starting material is broadly similar to the Parana´ basalts, in that they are tholeiitic and fairly evolved (Mg# of 35–60 compared to Mg# of 35–65 for the Parana´ basalts). One signiﬁcant diﬀerence in the starting composition between the experimental and Parana´ samples is the K2O abundance, which is markedly lower in the experimental starting material than the Parana´ basalts (<0·45 wt.% K2O as opposed to an average value of >1·3 wt.% K2O for the Parana´ basalts: Peate et al. 1992). Since K acts as an incompatible element throughout melting (neither amphibole or alkali-feldspar are residual solid phases in the melting reactions reported by Beard & Lofgren (1991), the abundance of K2O in the melt will depend on the concentration in the source, as well as on the degree of melting. Thus, the low K2O abundance of the experimental melts relative to the Parana´ rhyolites (<1 wt.% K2O vs. >3 wt.% K2O in the rhyolites) could reﬂect the diﬀerent K2O contents of their respective source rocks. The Parana´ rhyolites have high eruption temperatures of >1000(C (e.g. Garland et al. 1995), therefore the melt is required to be at equivalent or higher temperatures at the lowest observed silica content (w65 wt.% SiO2), assuming that temperatures are likely to decrease with increasing silica. The

Figure 4 SiO2 vs. temperature plot for the low-Ti Parana´ basalts and rhyolites (data from Piccirillo & Melﬁ 1988; Garland et al. 1995). Temperatures calculated by pyroxene thermometry (Kretz 1982), with errors. Both basalts and rhyolites lie mainly within the temperature range 950–1100(C. SiO2–temperature ﬁelds for the experimental data of Beard & Lofgren (1989) and Spulber & Rutherford (1983) are shown for comparison.

experiments of Spulber & Rutherford (1983) were carried out over a temperature range of 913–1050(C, but the temperatures drop rapidly from 52–60 wt.% SiO2, and at 65 wt.% SiO2 the temperature is <950(C (Fig. 4). This is at the lower end of the range of calculated temperatures for the rhyolites, whilst the experiments of Beard & Lofgren (1989, 1991) at 1000(C yield a melt with up to 66 wt.% SiO2. This makes the slightly higher PH2O experiments of Beard & Lofgren the preferred melting model for the Parana´ rhyolites. It is shown on Figure 3 that the Beard & Lofgren experiments (region 2) are oﬀset to consistently higher Al2O3 than the Parana´ rhyolites, but this can be accounted for by w10% plagioclase crystallisation. Further, this is consistent with the slightly higher CaO content in the experimental melts compared to the rhyolites, suggesting that the plagioclase removed was anorthite-rich, as opposed to albitic (plagioclase in the Parana´ basalts have the range An86–59: Bellieni et al. 1988). It must be borne in mind that whilst the basalt experiments have all been considered in terms of partial melting, the experiments are reversible, and therefore apply equally to crystallisation from a basalt melt of the same composition, under the same conditions of temperature and pressure (Spulber & Rutherford 1983). The Parana´ rhyolites could, in principle, be modelled either as partial melts of basaltic material, or as the result of crystallisation from a basalt. However, crystallisation processes are more energetically efﬁcient and thermal models for production of silicic melts via re-melting in the crust predict a secular change from silicic to maﬁc magmatism over time (Annen et al. 2006) – the opposite of that observed in most continental ﬂood basalt sequences. Thus, it seems likely that the low-Ti Parana´ rhyolites reﬂect extended fractionation of associated low-Ti basalts (Garland et al. 1995).

2.3. Comparison with other CFB rhyolites In the light of a strong link between the Parana´ rhyolites and experimental melts/fractionates from moderately waterundersaturated basalts, we now turn to rhyolites from other continental ﬂood basalt provinces world-wide to consider

56

SIMON TURNER AND TRACY RUSHMER

Figure 5 Al2O3 vs. FeO plot for rhyolites from continental ﬂood basalt provinces world-wide (see text for data sources). The trends marked 1 to 3 are best-ﬁt lines through the basalt melt ﬁelds shown in Figure 3, whilst (4) represents a best-ﬁt through the metapelite melt ﬁeld of Vielzeuf & Holloway (1988). For trend 2, the basalt melts of Beard & Lofgren (1989), temperatures in (C are shown for experiments carried out over the pressure range of 1–3 kb, with the low-temperature experiments resulting in relatively low FeO and low Al2O3 melts. The majority of the rhyolites lie close to the experimental trend of moderately dry basalts (1 and 2). Exceptions are the Columbia River rhyolite glasses, at the low temperature end of the wet basalt trend (3); and the Central Karoo silicics close to the metapelite melting ﬁeld (4). The vector for 10% plagioclase crystallisation is shown in the top right hand corner.

whether they have similar petrogenetic relations. The Karoo continental ﬂood basalt province has a highly voluminous rhyolite succession in the east of the province (Cleverly et al. 1984), where the Lebombo monocline is comprised almost entirely of rhyolite units (w35 000 km3). They can be divided into a northern high-Ti group, and a southern low-Ti group, with the lowermost ﬂow units (the Mkutshane Beds) constituting a separate low-Ti group with raised 87Sr/86Sr values, indicative of crustal contamination (Cleverly et al. 1984). There are volumetrically insigniﬁcant evolved ﬂows in the Central Karoo, which also have higher 87Sr/86Sr than the associated basalts, and have been modelled as crustal melts (Marsh & Eales 1984). Other continental ﬂood basalt provinces which contain silicic ﬂows, albeit in less impressive volumes than found in the Parana´–Etendeka and Karoo, are the Huronian (Jolly et al. 1992), the Keweenawan (Nicholson & Shirey 1990), and the Deccan provinces (Lightfoot et al. 1987). The British Tertiary province contains abundant evidence of intrusive silicic activity (e.g. Gamble et al. 1992), which is found on a smaller scale in the Tasmanian dolerites (McDougall 1962) and the Columbia River basalts (Lambert et al. 1989), the latter two being interpreted as in-situ diﬀerentiates from basic magma. The majority of erupted rhyolites in continental ﬂood basalt provinces share the same high temperatures as the Parana´ rhyolites (>1000(C: e.g. Cleverly et al. 1984; Green & Fitz 1993), and the majority have an anhydrous mineralogy, consisting of plagioclase, clinopyroxene, Fe–Ti oxides and rare quartz and alkali feldspar. These petrographic and thermal similarities between these rocks suggest that they were generated under broadly similar conditions of source composition, pressure and water content. In general, the rhyolites have eruption temperatures w200(C lower than their associated maﬁc rocks and such a temperature drop would result in 70–80 crystallisation, suﬃcient to yield a rhyolite from a basaltic precursor (Turner et al. 1992). These petrographic

and thermal similarities between these rhyolites suggest that they were generated under broadly similar conditions of source composition, pressure and water content. In order to make further comparison with the Parana´ rhyolites, and the experimental data reviewed above, the continental ﬂood basalt rhyolites are plotted on an Al2O3 versus FeO diagram (Fig. 5). The Lebombo rhyolites of the Karoo province, together with the Mkutshane Beds, plot very close to the Parana´ rhyolites, whilst the Huronian, Keweenawan and Deccan rhyolites plot on parallel trends adjacent to this same cluster (Fig. 5). A number of Deccan samples from Lightfoot et al. (1987) have been omitted from Figure 5 in order to avoid unnecessary cluttering for the following reasons: (i) samples with higher Al2O3 and FeO are interpreted as mixing with the associated basalts (noted for trace elements by Lightfoot et al. 1987); and (ii) samples with very high Al2O3 contents (16– 17 wt.% Al2O3) are thought to be the result of plagioclase accumulation (these samples are plagioclase-phyric, and have high Sr/Zr ratios: Lightfoot et al. 1987). The British Tertiary and Tasmanian silicic rocks plot at the low temperature end (lower Al2O3 and FeO) of the Beard & Lofgren (1989, 1991) experimental melts, consistent with the slower cooling of these rocks in intrusive bodies. By comparison with the composition of the experimental melts at pressures of 1–3 kb (Fig. 5), the British Tertiary rocks range in temperature from 900(C to 950(C, whilst the Tasmanian granophyres are marginally cooler at %900(C. Both the British Tertiary and the Tasmanian silicic rocks plot at the same Al2O3 contents as the Beard and Lofgren experimental melts, implying that subsequent plagioclase removal was not necessary to generate the observed magma composition. The Columbia River silicic glasses (found as interstitial melt pockets within the majority of basalt ﬂows, generally comprising w10% of the bulk ﬂow volume: Lambert et al. 1989) plot at the lowest FeO abundances (<2 wt.% FeO). They lie along the experimental melts

A-TYPE GRANITES AND RHYOLITES IN FLOOD BASALTS

57

Figure 6 SiO2 vs. TiO2 plot contrasting the high- and low-Ti Parana´ basalts. The low-Ti basalts show an initial increase in TiO2, whilst the high-Ti basalts show a consistent decrease in TiO2, indicating that titanomagnetite was crystallising from an early stage. Basalt data from Peate et al. (1992).

of the amphibolite source of Rapp et al. (1991), in which the melting reactions were ﬂuid-absent. The pressures at which the experiments were carried out are at higher pressures, ranging from 8 kb to 32 kb, and not as applicable to the lower pressures that would have been present for the emplacement conditions of the Columbia River Basalt province (Helz 1976). More applicable are the water-undersaturated basalt melting experiments of Holloway & Burnham (1972) in which PH2O =0·6Ptotal over the pressure range 5–8 kb. The fact that the Columbia River glasses lie along this possibly more water-enriched trend is indicative of the fact that water, along with slightly higher pressures when compared to those of Beard & Lofgren (1989, 1991) and Spulber & Rutherford (1983), plays an important role in changing phase stabilities and therefore FeO content of the melt. The only marked outlier from the rhyolite groups is the Central Karoo, which lies at high Al2O3 and moderate FeO relative to the rhyolites from other continental ﬂood basalt provinces. The rhyolites from the Central Karoo lie at very slightly lower Al2O3 contents than the high temperature experimental melts from metasediments of Vielzeuf & Holloway (1988), and are distinct from the other basalt melting trends. Marsh & Eales (1984) interpreted the Central Karoo silicic rocks as crustal melts based on a combination of composition and isotope ratios, and indeed melting of a two-mica metasedimentary rock at temperatures of R1150(C gives a melt of appropriate composition, although a slightly lower temperature melt, followed by 5–10% plagioclase removal would give a similar composition.

3. Crystallisation models and the lack of intermediate compositions The preceding discussion is based on reversible, equilibrium experiments which strictly only constrain equilibrium crystallisation. However, the diﬀerences between equilibrium and fractional crystallisation are generally trivial for major elements. The diﬀerence becomes more important for incompatible trace elements which are not the focus of the present paper, but can be used to distinguish between partial melting

and fractional crystallisation in some high silica magmas (e.g. Halliday et al. 1991). Therefore, it is now considered how fractional crystallisation processes may produce bimodal magmatic suites. Silica saturated basaltic magmas with a quartz– diorite composition, that typify continental ﬂood basalt provinces, tend to undergo continuous SiO2 enrichment with on-going crystallisation (e.g. Mahoney & Coﬃn 1997). By deﬁnition, intermediate compositions are rarely observed in bimodal suites, but this remains a weak argument against a fractionation origin for the silicic rocks (Brophy 1991; Turner et al. 1992; Bonnefol et al. 1995). In tholeiites, early ironenrichment is due to plagioclase (an iron-poor phase) dominating the early fractionating assemblage (Grove & Baker 1984), and FeO* abundance is then primarily controlled by Fe–Ti oxide crystallisation and removal. Due to the absence of silica from Fe–Ti oxides, removal of this phase signiﬁcantly increases SiO2 in the residual liquid. Within the low-Ti Parana´ basalts, TiO2 and Fe2O3 broadly increase with diﬀerentiation up to w56% SiO2, and the onset of Fe–Ti oxide crystallisation is then marked by an abrupt decrease in TiO2 and Fe2O3 with increasing SiO2, and the start of the silica gap (Fig. 6). Once Fe–Ti oxides begin to crystallise out, the calculated SiO2 content of the fractionating assemblage in the low-Ti basalts decreases from w48% to w42%. The onset of Fe–Ti oxide crystallisation is thus critical in determining the evolution of the silica content in the liquid. A gradual increase in silica, as demonstrated by the calc–alkaline series, is brought about because Fe–Ti oxides appear at high temperature, which also restricts iron-enrichment. The entry of Fe–Ti oxides into the crystallising assemblage is determined largely by temperature (Thy & Lofgren 1994) and fO2 (Snyder et al. 1993). High fO2 also leads to the early appearance of Fe–Ti oxides, particularly magnetite. Oxygen fugacities calculated for the Parana´ lavas using magnetite–ilmenite pairs generally lie along the QFM buﬀer (Bellieni et al. 1986, 1988), with the high-Ti magmas at marginally higher fO2 than the low-Ti magmas. This results in saturation of titanomagnetite earlier in the high-Ti magmas than in the low-Ti magmas. Recently, some workers have argued for the production of high silica magmas via the extraction of interstitial liquid from

58

SIMON TURNER AND TRACY RUSHMER

glass

amph

plag Figure 7 Photomicrograph of a Tasmanian dolerite containing residual silicic interstitial glass with a major element composition very similar to associated granophyric sills. Crossed polars; width=4 mm.

a cooling gabbroic crystal mush (Turner & Foden 1996; Marsh 2002; Bachmann & Bergantz 2004), providing a somewhat diﬀerent physical process for explaining the bimodal melt production. This is eﬀectively an equilibrium crystallisation process and can thus be directly linked to the experiments discussed above. A signiﬁcant observation is the occurrence of granophyre on various scales from interstitial segregations to larger sills and pods in continental ﬂood basalts and many maﬁc intrusions such as the Tasmanian dolerites (Fig. 7) and even basaltic lavas ﬂows (Martin & Sigmarsson 2007). In the study of Turner & Foden (1996) of the Mannum region in South Australia, the granophyre is similar in composition to many of the rhyolites in question, and overlaps the chemical and isotopic range of comagmatic A-type granites and volcanic rocks. This observation is a good example of the physical process of bimodal melt production suggested by Marsh (2002). In this model, it is the cooling of basaltic sills and gabbroic intrusions that drives instabilities and pressure gradients that allow concentration of silicic melts. Therefore, the coalescing of interstitial granophyric magma provides one possible petrogenetic model. Interstitial liquid in cumulates becomes enriched in components incompatible with the cumulus matrix of olivine, pyroxene and plagioclase. Intermediate compositions are not produced in this system because the components for intermediate compositions are removed by diﬀusion to become incorporated in overgrowths on the cumulus phases. Alkalis and incompatible elements will not readily become incorporated in these overgrowths and thus become enriched in a rhyolitic residua that develops into the interstitial granophyric phase. This A-type granitic liquid will have a lower solidus (800–725(C, Clemens et al. 1986) than the surrounding solid cumulate (cf. w1000(C), and thus will remain liquid for longer periods. It seems likely that such interstitial liquid, which is well documented from the Tasmanian Dolerites and other layered gabbroic plutons, should be capable of coalescing and/or gravitational separation. Compaction of the cumulate pile (e.g. Sparks et al. 1985) could force expulsion of this liquid to form sills at the top of the pile.

Interestingly Lambert et al. (1989) calculated that, if coalesced, the interstitial rhyolitic glass present in the Columbia River basalts would total some 10,000 km3 of rhyolite (although, in this case, no contemporaneous rhyolites were erupted).

4. Volumes and significance for crustal growth For a given starting mass of basalt (liquid or solid), both crystallisation of a basaltic liquid or partial melting of a solid basalt will produce the same relative volumes of maﬁc residue and a more silicious product. For example, the production of a siliceous melt with w70 SiO2 requires either >70% crystallisation of a basalt (e.g. Turner et al. 1992) or <30% partial melting of a basalt (e.g. Sisson et al. 2005). Thus, both the partial melting and crystallisation scenarios require that signiﬁcant volumes of basalt are emplaced into the crust during the formation of the rhyolites being discussed. Table 1 shows that all the of the provinces considered contain suﬃcient volumes of basalt to account for the rhyolites. In this context, it has long been noted that continental ﬂood basalts contribute signiﬁcant volumes (w810 1 km3/yr) of new maﬁc material to the continental crust (e.g. Hawkesworth et al. 1999). However, the associated volumes of rhyolite are also non-trivial. Table 1 lists the approximate volumes of rhyolite associated with some of the major Phanerozoic continental ﬂood basalts, and the calculated rate of addition of this silicic material (w410 4 km3/yr) to the continental crust compares favourably, for example, with an estimate of 4·410 4 km3/yr for the Central Andes (Crisp 1984).

5. Summary and conclusions It has long been inferred that the major site of production of silicic continental crust is active magmatic convergent margins. However, the importance of silica-rich magmas that are produced during diﬀerentiation of continental ﬂood basalts may be underestimated. The present authors suggest that the

A-TYPE GRANITES AND RHYOLITES IN FLOOD BASALTS

rhyolites that are found associated with continental ﬂood basalts (Parana´ in particular) may be formed by fractional crystallisation during cooling. Geochemical data from the natural rhyolites and experimental studies of melts formed from basaltic starting material (performed under low to moderate pressures at PH2O
6. Acknowledgements It is a pleasure to thank Francis Garland, John Foden and Chris Hawkesworth for many interesting discussions on this topic over the years. We also thank Eric Christiansen and John Shervais for detailed reviews, and Carol Frost for editorial handling. Simon Turner was funded by an ARC Federation Fellowship.

7. References Annen, C., Blundy, J. D. & Sparks, R. S. J. 2006. The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology 47, 505–39. Bachmann, O. & Bergantz, G. W. 2004. On the origin of crystal-poor rhyolites: extracted from batholithic crystal mushes. Journal of Petrology 45, 1565–82. Beard, J. S. & Lofgren, G. E. 1989. Eﬀects of water on the composition of partial melts of greenstone and amphibolite. Science 244, 195–7. Beard, J. S. & Lofgren, G. E. 1991. Dehydration melting and water-saturated melting of basaltic and andesitic greenstones and amphibolites at 1·3 and 6·9 kbar. Journal of Petrology 32, 365– 401. Bellieni, G., Comin-Chiaramonti, P., Marques, L. S., Melﬁ, A. J., Nardy, A. J. R., Papatrechas, C., Piccirillo, E. M., Roisenberg, A. & Stolfa, D. 1986. Petrogenetic aspects of acid and basaltic lavas from the Parana´ plateau (Brazil): geological, mineralogical and petrochemical relationships. Journal of Petrology 27, 915–44. Bellieni, G., Piccirillo, E. M., Comin-Chiaramonti, P., Melﬁ, A. J. & Da Roit, P. 1988. Mineral chemistry of continental stratoid volcanics and related intrusives from the Parana´ Basin (Brazil). In Piccirillo, E. M. & Melﬁ, A. J. (eds) The Mesozoic flood volcanics of the Parana´ Basin: petrogenetic and geophysical aspects, 73–92. Sa˜o Paulo: IAG–USP Press. Bonnefol, C. C., Provost, A. & Albarede, F. 1995. The ‘Daly gap’ as a magmatic catastrophe. Nature 378, 270–2. Brophy, J. G. 1991. Composition gaps, critical crystallinity, and fractional crystallization in orogenic (calc–alkaline) magmatic systems. Contributions to Mineralogy and Petrology 109, 173–82. Collins, W. J., Beams, S. D., White, A. J. R. & Chappell, B. W. 1982. Nature and origin of A-type granites with particular reference to south-eastern Australia. Contributions to Mineralogy and Petrology 80, 189–200. Clemens, J. D., Holloway, J. R. & White, A. J. R. 1986. Origin of A-type granite: Experimental constraints. American Mineralogist 71, 317–24. Clemens, J. D. & Vielzeuf, D. 1987. Constraints on melting and magma production in the crust. Earth and Planetary Science Letters 86, 287–306. Cleverly, R. W., Betton, P. J. & Bristow, J. W. 1984. Geochemistry and petrogenesis of the Lebombo rhyolites. Geological Society of South Africa Special Publication 13, 171–94. Creaser, R. A. & White, A. J. R. 1991. Yardea Dacite – Large volume, high-temperature felsic volcanism from the Middle Proterozoic of South Australia. Geology 19, 48–51. Crisp, J. A. 1984. Rates of magma emplacement and volcanic output. Journal of Volcanology and Geothermal Research 20, 177–211. Eby, G. N. 1990. The A-type granitoids: A review of their occurrence and chemical characteristics and speculations on their petrogenesis. Lithos 26, 115–34. Ekren, E. B., McIntyre, D. H. & Bennett, E. H. 1984. High temperature, large volume, lavalike ﬂow tuﬀs without calderas in southeastern Idaho. US Geological Survey Professional Paper 1272, 1–73.

59

Ewart, A., Marsh, J. S., Milner, S. C., Duncan, A. R., Kamber, B. S. & Armstrong, R. A. 2004. Petrology and geochemistry of early Cretaceous bimodal continental ﬂood volcanism of the NW Etendeka, Namibia. Part 2: Characteristics and petrogenesis of the high-Ti latite and high-Ti and low-Ti voluminous quartz latite eruptives. Journal of Petrology 45, 107–38. Frost, C. D. & Frost, B. R. 1997. Reduced rapakivi-type granites: the tholeiite connection. Geology 25, 647–50. Gamble, J. A., Meighan, I. G. & McCormick, A. G. 1992. The petrogenesis of Tertiary microgranites and granophyres from the Slieve Gullion central complex, NE Ireland. Journal of the Geological Society, London 149, 93–106. Garland, F., Hawkesworth, C. J. & Mantovani, M. S. M. 1995. Description and petrogenesis of the Parana´ rhyolites. Journal of Petrology 36, 1193–227. Green, J. C. & Fitz, T. J. 1993. Extensive felsic lavas and rheoignimbrites in the Keweenawan Midcontinent Rift plateau volcanics, Minnesota: petrographic and ﬁeld recognition. Journal of Volcanology and Geothermal Research 54, 177–96. Grove, T. L. & Baker, M. B. 1984. Phase equilibrium controls on the tholeiitic versus calc–alkaline diﬀerentiation trends. Journal of Geophysical Research 89, 3253–74. Halliday, A. N., Davidson, J. P., Hildreth, W. & Holden, P. 1991. Modelling the petrogenesis of high Rb/Sr silicic magmas. Chemical Geology 92, 107–14. Hawkesworth, C., Kelley, S., Turner, S., Le Roex, A. & Storey, B. 1999. Mantle processes during Gondwana break-up and dispersal. Journal of African Earth Sciences 28, 239–61. Helz, R. T. 1976. Phase relations of basalts in their melting ranges at PH2O =5 kbar. Part II. Melt compositions. Journal of Petrology 17, 139–93. Henry, C. D., Price, J. G., Rubin, J. N., Parker, D. F., Wolﬀ, J. A., Self, S., Franklin, R. & Barker, D. S. 1988. Widespread, lavalike silicic volcanic rocks of Trans-Pecos Texas. Geology 16, 509–12. Hergt, J. M. 1987. The origin and evolution of the Tasmanian dolerites. PhD Thesis. The Australian National University, Canberra. Holloway, J. R. & Burnham, C. W. 1972. Melting relationships of basalt with equilibrium water pressure less than total pressure. Journal of Petrology 13, 1–29. Holtz, F. & Johannes, W. 1994. Maximum and minimum water contents of granitic melts: implications for chemical and physical properties of ascending magmas. Lithos 32, 149–59. Huppert, H. E. & Sparks, R. S. J. 1988. The generation of granitic magmas by intrusion of basalt into continental crust. Journal of Petrology 29, 599–624. Jolly, W. T., Dickin, A. P. & Wu, T.-W. 1992. Geochemical stratigraphy of the Huronian continental volcanics at Thessalon, Ontario: contributions of two-stage crustal fusion. Contributions to Mineralogy and Petrology 110, 411–28. Kretz, R. 1982. Transfer and Exchange equilibria in a portion of the pyroxene quadrilateral as deduced from natural and experimental data. Geochimica et Cosmochimica Acta 46, 411–21. Lambert, R. St. J., Marsh, I. K. & Chamberlain, V. E. 1989. The occurrence of interstitial granite-glass in all formations of the Columbia River Basalt Group and its petrogenetic implications. Geological Society of America Special Paper 239, 321–32. Leeman, W. P. 1982. Rhyolites of the Snake River Plain–Yellowstone Plateau Province, Idaho and Wyoming: a summary of petrogenetic models. In Bonnischen, B. & Breckenridge, R. M. (eds) Cenozoic Geology of Idaho. Idaho Bureau of Mines and Geology Bulletin 26, 63–6. Lightfoot, P. C., Hawkesworth, C. J. & Sethna, S. F. 1987. Petrogenesis of rhyolites and trachytes from the Deccan Trap: Sr, Nd and Pb isotope and trace element evidence. Contributions to Mineralogy and Petrology 95, 44–54. McCurry, M., Hayden, K. P., Morse, L. H. & Mertzman, S. 2008. Genesis of post-hotspot, A-type rhyolite of the Eastern Snake River Plain volcanic ﬁeld by extreme fractional crystallization of olivine tholeiite. Bulletin of Volcanology 70, 361–83. McDougall, I. 1962. Diﬀerentiation of the Tasmanian dolerites: Red Hill dolerite-granophyre association. Geological Society of America Bulletin 73, 279–316. Mahoney, J. J. & Coﬃn, M. F. (eds) 1997. Large igneous provinces: continental, oceanic, and planetary flood volcanism. Geophysical Monograph 100, 438 pp. Washington, DC: American Geophysical Union. Marsh, B. D. 2002. On bimodal diﬀerentiation by solidiﬁcation front instability in basaltic magmas, part 1: Basic mechanics. Geochimica et Cosmochimica Acta 66, 2211–29.

60

SIMON TURNER AND TRACY RUSHMER

Marsh, J. S. & Eales, H. V. 1984. The chemistry and petrogenesis of igneous rocks of the Karoo Central Area, Southern Africa. Geological Society of South Africa Special Publication 13, 27–69. Martin, E. & Sigmarsson, O. 2007. Low-pressure diﬀerentiation of tholeiitic lavas as recorded in segregation veins from Reykjanes (Iceland), Lanzarote (Canary Islands) and Masaya (Nicaragua). Contributions to Mineralogy and Petrology 154, 559–73. Milner, S. C. & Ewart, A. 1989. The geology of the Goboboseb Mountain volcanics and their relationship to the Messum Complex, Namibia. Communications of the Geological Survey of Namibia 5, 31–40. Morse, S. A. 1980. Basalts and phase diagrams: an introduction to the quantitative use of phase diagrams in igneous rocks. Berlin, Heidelberg, New York: Springer-Verlag. 493 pp. Nicholson, S. W. & Shirey, S. B. 1990. Midcontinent Rift volcanism in the Lake Superior region: Sr, Nd, and Pb isotopic evidence for a mantle plume origin. Journal of Geophysical Research 95, 10851– 68. Patin˜o Douce, A. E. & Johnston, A. D. 1991. Phase equilibria and melt productivity in the pelitic system: implications for the origin of peraluminous granitoids and aluminous granulites. Contributions to Mineralogy and Petrology 107, 202–18. Pearce, J. A., Harris, N. B. W. & Tindle, A. G. 1984. Trace element discrimination diagrams for the tectonic interpretation of granitic rocks. Journal of Petrology 25, 956–83. Peate, D. W., Hawkesworth, C. J. & Mantovani, M. S. M. 1992. Chemical stratigraphy of the Parana´ lavas, South America: classiﬁcation of magma types and their spatial distribution. Bulletin of Volcanology 55, 119–39. Piccirillo, E. M., Bellieni, G., Cavazzini, G., Comin-Chiaramonti, P., Petrini, R., Melﬁ, A. J., Pinese, J. P. P., Zantadeschi, P. & De Min, A. 1987. Bimodal ﬁssural volcanic suites from the Parana´ Basin (Brazil): K–Ar age, Sr-isotopes and geochemistry. Geochemica Brasiliensis 1, 53–69. Piccirillo, E. M. & Melﬁ, A. J. 1988. The Mesozoic flood volcanism of the Parana´ Basin: petrogenetic and geophysical aspects. Sa˜o Paulo: IAG–USP Press. 600pp. Rapp, R. P., Watson, E. B. & Miller, C. F. 1991. Partial melting of amphibolite/eclogite and the origin of Archaean trondhjemites and tonalites. Precambrian Research 51, 1–25. Rapp, R. P. & Watson, E. B. 1995. Dehydration melting of metabasalt at 8–32 kbar: implications for continental growth and crustmantle recycling. Journal of Petrology 36, 891–931. Rushmer, T. 1991. Partial melting of two amphibolites: contrasting experimental results under ﬂuid-absent conditions. Contributions to Mineralogy and Petrology 107, 41–59. Rutter, M. J. & Wyllie, P. J. 1988. Melting of vapour-absent at 10 kb to simulate dehydration-melting in the deep crust. Nature 331, 159–60.

Sisson, T. W., Ratajeski, K., Hankins, W. B. & Glazner, A. F. 2005. Voluminous granitic magmas from common basaltic sources. Contributions to Mineralogy and Petrology 148, 635–61. Skjerlie, K. P. & Johnston, A. D. 1993. Fluid-absent melting behaviour of an F-rich tonalitic gneiss at mid-crustal pressures: implications for the generation of anorogenic granites. Journal of Petrology 34, 785–815. Snyder, D., Carmichael, I. S. E. & Wiebe, R. A. 1993. Experimental study of liquid evolution in an Fe-rich, layered maﬁc intrusion: constraints of Fe–Ti oxide precipitation on the T–fO2 and T–p paths of tholeiitic magmas. Contributions to Mineralogy and Petrology 113, 73–86. Sparks, R. S. J., Huppert, H. E., Kerr, R. C., McKenzie, D. P. & Tait, S. R. 1985. Post cumulus processes in layered intrusions. Geological Magazine 122, 555–68. Spulber, S. D. & Rutherford, M. J. 1983. The origin of rhyolite and plagiogranite in oceanic crust: an experimental study. Journal of Petrology 24, 1–25. Sylvester, P. J. 1989. Post-collisional alkaline granites. Journal of Geology 97, 261–80. Thy, P., Beard, J. S. & Lofgren, J. E. 1990. Experimental constraints on the origin of Icelandic rhyolites. Journal of Geology 98, 417–21. Thy, P. & Lofgren, G. E. 1994. Experimental constraints on the low-pressure evolution of transitional and mildly alkalic basalts: the eﬀect of Fe–Ti oxide minerals and the origin of basaltic andesites. Contributions to Mineralogy and Petrology 116, 340–51. Turner, S. P. 1996. Petrogenesis of the late-Delamerian gabbroic complex at Black Hill, South Australia: implications for convective thinning of the lithospheric mantle. Mineralogy and Petrology, 56, 51–98. Turner, S. P., Foden, J. D. & Morrison, R. S. 1992. Derivation of some A-type magmas by fractionation of basaltic magma and an example from the Padthaway Ridge, South Australia. Lithos 28, 151–71. Turner, S. P. & Foden, J. D. 1996. Magma mingling in lateDelamerian A-type granites at Mannum, South Australia. Mineralogy and Petrology 56, 147–69. Vielzeuf, D. & Holloway, J. R. 1988. Experimental determination of the ﬂuid-absent melting relations in the pelitic system. Contributions to Mineralogy and Petrology 98, 257–76. Walsh, J. N., Beckinsdale, R. D., Skelhorn, P. R. & Thorpe, R. S. 1979. Geochemistry and petrogenesis of Tertiary granitic rocks from the Island of Mull, northwest Scotland. Contributions to Mineralogy and Petrology 71, 99–116. Whitaker, M. L., Nekvasil, H., Lindsley, D. H. & McCurry, M. 2008. Can crystallization of olivine tholeiite give rise to potassic rhyolites? – An experimental investigation. Bulletin of Volcanology 70, 417–34.

MS received 27 December 2007. Accepted for publication 13 May 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 61–76, 2010 (for 2009)

Structural controls on melt segregation and migration related to the formation of the diapiric Schwerin Fold in the contact aureole of the Bushveld Complex, South Africa Luke Longridge1, Roger L. Gibson1 and Paul A. M. Nex1,2 1

School of Geosciences, University of the Witwatersrand, Johannesburg, PVT Bag 3, Wits, 2050, South Africa Email: [email protected]

2

Umbono Group of Companies, Isle of Houghton, Johannesburg, South Africa

ABSTRACT: Partial melting of metapelitic rocks beneath the maﬁc–ultramaﬁc Rustenburg Layered Suite of the Bushveld Complex in the vicinity of the periclinal Schwerin Fold resulted in a structurally controlled distribution of granitic leucosomes in the upper metamorphic aureole. In the core of the pericline, subvertical structures facilitated the rise of buoyant leucosome through the aureole towards the contact with the Bushveld Complex, with leucosomes accumulating in en-echelon tension gashes. In a subhorizontal syn-metamorphic shear zone to the southeast of the pericline, leucosomes accumulated in subhorizontal dilational structural sites. The kinematics of this shear zone are consistent with slumping of material oﬀ the southeastern limb of the rising Schwerin pericline. The syndeformational timing of leucosome emplacement supports a syn-intrusive, densitydriven origin for the Schwerin Fold. Modelling of the cooling of the Rustenburg Layered Suite and heating of the ﬂoor rocks using a multiple intrusion model indicates that temperatures above the solidus were maintained for >600,000 years up to 300 m from the contact, in agreement with rheological modelling of ﬂoor-rock diapirs that indicate growth rates on the order of 8 mm/year for the Schwerin Fold. KEY WORDS:

diapiric fold, granite leucosomes, thermal modelling

The role of deformation in the initial segregation, migration and accumulation of leucosomes has been investigated in a wide variety of regional metamorphic terrains (e.g. Brown 1994 and references therein; Brown & Solar 1998; Kisters et al. 1998). Additionally, studies have demonstrated the weakening eﬀect of melt on the rheology of rocks (e.g. Bai et al. 1997; Gleason et al. 1999; Holyoke & Rushmer 2002). Although many studies exist that demonstrate the relationship between deformation and melting in an orogenic context, few studies exist where melting and deformation relationships have been examined in contact metamorphic aureoles (e.g. Rosenberg & Berger 2000; Vernon et al. 2003). In the northeastern parts of the contact aureole to the Bushveld Complex, South Africa, a number of kilometre-scale periclinal fold structures are found in the upper amphibolite- to granulite-grade ﬂoor rocks underlying the Bushveld Complex. The Schwerin Fold is one of these periclinal structures developed in migmatitic rocks of the upper Transvaal Supergroup. Here, the interplay of deformation and anatexis and, particularly, the positive feedback loop between deformation-enhanced melt accumulation and migration, and the rheological consequences of deformation in partially molten rocks, can be investigated.

buoyancy of low-density granitic magma is a primary driving force for the ascent of the magma, fault systems and shear zones are needed as conduits. In addition to gravity-driven ascent of buoyant magma, non-hydrostatic stresses and pressure gradients may also drive melt into, and along structures. Strain localisation in migmatite terrains is a key factor in the accumulation, segregation and transport of granitic melt, and a positive feedback exists between deformation-initiated melt mobilisation, melt accumulation promoting ductile deformation processes and enhancing bulk strain partitioning, and the buoyant rise of segregated melts (Kisters et al. 1998). In orogenic terrains, where heating of rocks to upper amphibolite to granulite facies conditions is generally accompanied by progressive deformation, this deformation can result in the formation of low-pressure sites where melt can preferentially accumulate. These low-pressure sites can include boudin necks (Vernon et al. 2003), extensional shear bands (Brown et al. 1995), shear zones (Bhadra et al. 2007), and fold axial planes (Hand & Dirks 1992). These low-pressure sites also play a key role in the segregation of maﬁc selvedges from leucosome (Sawyer 1991). This melt accumulation leads to a localised increase in the ductility of the rocks in the vicinity of the melt, which in turn increases strain partitioning in the vicinity of the melt.

1. Melting sites and melt migration in metamorphic terranes

2. Regional geological setting and previous work

The importance of deformation in the migration and ascent of granitic magmas in orogens has been noted repeatedly (e.g. Brown 1994, 2005; Kisters et al. 1998) and, although the

The intrusion of the w8 km-thick, 2·06 Ga (Walraven et al. 1990) sill-like maﬁc-ultramaﬁc Rustenburg Layered Suite (RLS) of the Bushveld Complex (Fig. 1) into the sedimentary

2009 The Royal Society of Edinburgh. doi:10·1017/S1755691009016119

62

LUKE LONGRIDGE ET AL.

Figure 1 Geological map of the northeastern limb of the Bushveld Complex, showing locations of the Schwerin Fold (A) and other fold structures: B=Katkloof; C=Phosiri; D=Phepane; E=Steelpoort; F=Derde Gelid. After 1:250 000 Geological Series sheets 2429B (Chuniespoort) and 2430A (Wolkberg) (Schwellnus et al. 1962).

rocks of the Transvaal Supergroup formed a metamorphic aureole in the ﬂoor rocks that extends up to tens of kilometres from the RLS contact. Since the RLS intruded horizontally, the term ‘upper aureole’ is used to describe the highest-grade rocks that lie immediately below the contact with the RLS. In the upper aureole, metamorphic grade reached granulite grade (indicated by the presence of metamorphic orthopyroxene locally in rocks proximal to the RLS contact; Uken 1998; Johnson et al. 2003), and migmatites formed in pelitic and psammitic rocks. Mapping of the ﬂoor rocks to the RLS in the northeastern Bushveld Complex has revealed a number of domes and periclines within both these high-grade migmatites and the lower-grade unmigmatised metapelites (Fig. 1). Early workers suggested that these structures formed as a result of regional compression and modiﬁcation of pre-Bushveld folds during the intrusion of the RLS (Sharpe & Chadwick 1982; du Plessis and Walraven 1990; Hartzer 1995), but more recent work (Uken & Watkeys 1997; Uken 1998; Johnson et al. 2004) has suggested that these structures are the result of gravityinduced diapiric rise of low-density argillaceous, arenaceous and calcareous rocks into the dense maﬁc magmas of the RLS (Uken & Watkeys 1997), a mechanism initially proposed by Button (1978). The Schwerin Fold (located w60 km NNW of

the town of Burgersfort; Fig. 1) is one such pericline and is developed in rocks of the upper Pretoria Group (Transvaal Supergroup; Fig. 2). Contact metamorphic eﬀects beneath the RLS in the northeastern Bushveld Complex extend throughout the Pretoria Group, with temperatures of w450(C in rocks of the lower Pretoria Group (Waters & Lovegrove 2002), and up to w750(C near the contact with the RLS (Johnson et al. 2003, 2004). Bulk compositional variation, as well as metamorphic grade, controls mineral assemblages in the aureole. A regional study of anatectic migmatites in the eastern Bushveld Complex aureole by Johnson et al. (2003) emphasised the role of deformation in the migration and ascent of melts. They noted that, throughout the upper aureole, migmatites occur in rocks that are gently folded on a m-scale, with a moderate to weak foliation, and that are locally more intensely strained (particularly in the vicinity of the km-scale periclines). In the zone closest to the contact with the RLS, Johnson et al. (2003) described migmatites associated with a variety of structural features. These include quartz-sillimanite veins, approximately parallel to the subhorizontal bedding in the area, which are folded into cm- to dm-scale open to tight, locally ptygmatic folds, and which contain a foliation similar to that in the host

MELT SEGREGATION AND MIGRATION IN THE SCHWERIN FOLD

63

Figure 2 Stratigraphy of the Transvaal Supergroup showing the detailed stratigraphy and lithologies of the Pretoria Group in the study area. Formations which dominate the study area are shaded in grey. Modiﬁed after Eriksson et al. (1993).

rocks. They interpreted these features as channels for H2Oand volatile-rich melts. Also noted were higher-strain domains commonly associated with discordant leucosomes, that are commonly located above the small antiformal closures of folded quartz-sillimanite veins. The signiﬁcance of this contemporaneous relationship between melting and deformation is explored further below.

3. Geology of the Schwerin Fold The Schwerin Fold (Figs 1, 3) is a kilometre-scale, upright to steeply inclined, gently- to moderately-plunging periclinal structure that is developed in the arenaceous to argillaceous rocks (with minor pyroclastic and calcareous rocks) of the Pretoria Group (uppermost Transvaal Supergroup; Fig. 2). The contact with the RLS is at the level of the Vermont Formation. The Schwerin Fold has a maximum wavelength of w8 km and an amplitude of w5 km, but is disharmonic, and the interlimb angle decreases to that of a tight fold in the stratigraphically-highest layers closest to the RLS. The fold is asymmetric, with steeper dips (70–80() on the western limb and shallower dips (20–40() on the eastern limb (Fig. 3). The regional dip of the rocks in the area is towards the RLS (w25(SW), subparallel to the magmatic layering in the RLS. The Schwerin Fold diﬀers from other periclines in the ﬂoor to the RLS in its asymmetry, and in that it has a curved axial trace, which is perpendicular to the contact with the RLS at the contact but curves to subparallel to the RLS strike away from the contact. However, like the other periclines in the RLS aureole (Uken & Watkeys 1997; Uken 1998), it has gentle to open synclinal areas adjacent to both limbs. For simplicity of

structural analysis, the Schwerin Fold has been divided into three sub-domains.

3.1. Domain 1 – metasediments in the contact aureole away from the Schwerin Fold At a distance of 3–6 km from the contact with the RLS, in the NE of the study area, pelitic rocks of the Lydenburg Member (Silverton Formation; Fig. 2) dip shallowly (w25() SW. The rocks retain original sedimentary compositional layering, deﬁned by layers which are more or less biotite-rich (Fig. 4A). A moderately SE-dipping (w35(), moderate to weak foliation deﬁned by biotite occurs oblique to bedding (Fig. 4B). The rocks are not migmatised and contain the assemblage andalusite–biotite–cordierite–quartz–ilmenite/magnetite. The matrix consists of ﬁne (0·01–0·2 mm) quartz, biotite and ilmenite/magnetite. Andalusite porphyroblasts are generally stubby to elongate, 1–5 cm long, chiastolitic, and lie with their longest dimensions within the bedding plane (indicating a compositional control). Andalusite contains graphite and small (0·05–0·1 mm) muscovite and quartz inclusions towards the core, and larger (0·1–0·2 mm) quartz and biotite inclusions near the rim. Cordierite porphyroblasts are 0·5–5 mm in size, and some cordierite porphyroblasts are texturally zoned, with an inclusion-rich core and an inclusion-free rim. Cordierite contains ﬁne (0·01–0·1mm) inclusions of quartz, biotite and ilmenite/magnetite, which are aligned parallel to the foliation observed in the matrix. Andalusite has inclusions aligned parallel to the foliation, although this is less distinctive owing to the generally inclusion-free nature of andalusite relative to cordierite. No pressure shadows exist adjacent to either andalusite or cordierite (which, together with alignment of inclusions, supports post-tectonic growth). This contrasts with

64

LUKE LONGRIDGE ET AL.

Figure 3 Geological map of the Schwerin Fold, showing the lower limit of migmatite development in the contact aureole. Note the change in strike of the axial surface and the steeper dip of strata on the western limb. Modiﬁed after Uken (1998).

textures further from the RLS contact reported by Uken (1998), which show ﬂattening of the bedding-parallel foliation around andalusite porphyroblasts.

3.2. Domain 2 – metasediments from the Schwerin Fold The second domain includes metapelites (Lydenburg Member) in the core of the Schwerin Fold, as well as the quartzites (Magaliesberg Formation) closer to the contact with the RLS (Fig. 3). Closest to the contact with the RLS, the Magaliesberg Formation quartzite forms a prominent ridge. This quartzite describes a moderately SW-plunging anticlinal fold with a 3 km wavelength. Bedding dip varies from w70(WSW to vertical on the western limb of the fold to w40(S to SSE on the southeastern limb of the fold. The quartzites are well bedded, and retain numerous original sedimentary features, including trough cross-bedding and ripple marks. Crossbedding conﬁrms that the rocks have not been overturned. The rocks are coarsely recrystallised and 2–3 cm long euhedral quartz crystals are found ﬁlling vugs in the quartzite, consistent with hydrothermal ﬂuid activity. In places on the western limb a weak foliation is deﬁned by elongate quartz grains. This foliation is slightly oblique to bedding, striking 010( and dipping steeply at 70(W. Gentle to open folds, with a 2–5 m wavelength and 0·5 m to 1 m amplitude are present in the quartzites, and plunge steeply NW. The core of the Schwerin Fold is deﬁned by a ridge of pelitic to semi-pelitic rocks of the Lydenburg Member (Figs 2, 3) and calc-silicates and volcanic breccias of the Machadodorp Member (Fig. 2). Bedding is deﬁned by mm- to cm-thick

siliceous layers and sedimentary structures (graded, ﬂaser and rare trough cross-bedding) that indicate a right-way-up sequence. Bedding is generally moderately dipping, but has a variable orientation caused by dm- to m-scale open to tight folds. A penetrative subvertical foliation is present which, at distances of >2 km from the contact with the RLS, is deﬁned by biotite. The northeastern parts of Domain 2 straddle the lower limit of migmatite development, and contain the assemblage andalusite–biotite–cordierite–plagioclase–muscovite–quartz in lower-grade metapelites, and andalusite–biotite–cordierite– plagioclase–muscovite–ﬁbrolite–quartz in higher-grade samples. Biotite occurs as 0·5–1 mm grains with 0·02–0·1 mm inclusions of quartz, although it is generally inclusion-free. It is locally retrogressed to chlorite. Texturally-late muscovite is present as large (>1 mm) vermicular grains, which overgrow the foliation. Cordierite is 0·5–1 mm in size, and contains ﬁne (0·01 mm) inclusions of quartz and coarser (0·1 mm) inclusions of biotite (Fig. 4H). It is commonly texturally zoned, with inclusion-free rims surrounding inclusion-rich cores, and may show sector twinning. Cordierite porphyroblasts are commonly lensoid and aligned with the foliation (indicating syntectonic growth). In higher-grade samples, ﬁbrolite is developed around biotite, and appears to form at the expense of biotite (Fig. 4G). Where ﬁbrolite is developed, it is not distributed throughout the sample, but occurs as discrete ‘seams’ within the sample, along which it replaces biotite. Locally, 1–3 cm-long andalusite porphyroblasts are aligned parallel to the biotite and ﬁbrolite foliations, supporting

MELT SEGREGATION AND MIGRATION IN THE SCHWERIN FOLD

syn-tectonic growth, although textural evidence indicates that ﬁbrolite postdates andalusite. A thin garnet-bearing horizon below the migmatite zone contains the assemblage garnet–biotite–cordierite–muscovite– ilmenite–quartz, with 0·2–1 mm subhedral garnet porphyroblasts preferentially developed in ilmenite-rich beds. Garnet contains 0·01–0·02 mm inclusions of quartz and ilmenite, and the moderate biotite foliation may bend gently around the garnet porphyroblasts, although generally it does not wrap around garnet, suggesting late- to post-tectonic growth of the latter. Within w2 km of the outcrop of the RLS contact, the pelitic rocks are cut by coarse-grained to pegmatitic quartz–feldspar– ﬁbrolite–tourmaline–muscovite leucosomes, which may contain peritectic cordierite (see also Johnson et al. 2003). As the contact is approached, a progressive increase in the volume of leucosome is seen. In the lowermost migmatite zone, the leucosomes develop within metapelites adjacent to quartzsillimanite veins (originally more psammitic beds; Johnson et al. 2003) as 1–5 cm long, 0·5–2 cm wide veins (Fig. 4D) but, with decreasing distance from the contact, they develop into tension gashes up to 30 cm in length and w5 cm wide (Fig. 4E). As the volume of leucosome increases, the leucosomes become more pod-like, and cross-cut the pelite bedding, forming an interconnected network (Fig. 4F). The leucosomes are coarsely crystalline (quartz grains w5 mm) to pegmatoidal (quartz grains w20 mm). Leucosomes with the ﬁnest grain size retain a relict ﬁbrolite foliation within them, with a similar orientation to that observed in the pelites, but this foliation is absent in the coarser pegmatites. Granophyric textures are observed; these commonly overgrow ﬁbrolite aggregates, which also occur as inclusions in quartz. Although the leucosomes appear in structurally controlled sites, there is no evidence that deformation continued following leucosome crystallisation. Minerals within the leucosomes are unstrained, and leucosomes show no penetrative fabric. The modal proportion of ﬁbrolite increases in tandem with the volume of leucosome towards the RLS contact. Textural evidence suggests that ﬁbrolite development was at the expense of biotite, rather than andalusite, and that ﬁbrolite-rich seams correspond to pathways recording the passage of an H2O-rich volatile phase, which may have been responsible for the ﬂuid-ﬂuxed incongruent melting of biotite (Johnson et al. 2003). Johnson et al. (2003) suggested that the reaction Pl+Bt+Qtz=Kfs+Crd+L is the most likely melt-producing reaction for typical Lydenburg Member metapelite compositions, although melt-producing reactions vary with temperature and XMg, and an inﬂux of H2O may have generated the congruent melting reaction Kfs+Pl+Bt+Crd+Qtz+H2O=L. A maﬁc sill is found at the base of the Lydenburg Member in both the migmatites and lower grade rocks and (like the stratigraphy) is cut by the isograds in the aureole (Fig. 3). This sill has no obvious chill margin, but ﬁeld evidence (textural and mineralogical variation) suggests that it may be a composite sill composed of at least two phases. It is metamorphosed, with an assemblage tremolite–actinolite–plagioclase–quartz indicating signiﬁcant rehydration, and does not preserve any original igneous textures.

3.3. Domain 3 – migmatitic shear zone east of the Schwerin Fold Domain 3 is found in the southeastern part of the study area, on the southeastern limb of the syncline ﬂanking the Schwerin Fold (Fig. 3). Here the uppermost rocks (Silverton Formation) contain a 50 m to 100 m-wide subhorizontal to gently SWdipping bedding-parallel shear zone (Fig. 3). Below the shear zone, metapelites preserve compositional banding as alternat-

65

ing biotite-rich and quartz-feldspar-rich layers, or as staurolite-rich layers. In less competent lithologies in the shear zone, bedding appears transposed, with rootless tight to isoclinal cm-scale folds of either quartz–feldspar or biotite-rich layers, indicating remnant bedding (Fig. 5A). Assemblages in more biotite-rich layers are generally quartz–biotite–plagioclase feldspar, with some perthitic K-feldspar, cordierite and muscovite. Generally the rock consists of ﬁne (0·02–0·1 mm) quartz, plagioclase and biotite with a strong foliation, and 0·5–1 mm vermicular muscovite overgrowing the foliation. Cordierite is 0·5–1 mm in size, and contains 0·01 mm inclusions of biotite and quartz. Andalusite is rare, and no ﬁbrolite is developed in these rocks. A shallow WNW-dipping foliation, deﬁned by biotite, wraps around low strain lenses that contain a relict foliation with an obliquity of 15( to 70( relative to the main foliation (Fig. 5B). Pressure shadows around these low strain lenses contain slightly coarser (0·2–0·4 mm) quartz and feldspar (Fig. 5B). No change in assemblage is observed between these low strain lenses and the foliated matrix. The shear zone is characterised by open, tight and isoclinal m-scale folding (Fig. 5A, C), with fold vergence typically towards the S, SW or W, and an axial planar foliation, deﬁned by biotite, which typically dips shallowly WNW (Fig. 5C). Locally, mullions are present in fold hinges. Listric extensional shear bands (ESBs) are widespread, and are ﬁlled with coarse to pegmatoidal quartz–feldspar–tourmaline–muscovite leucosomes, containing 0·5–5 cm biotite schlieren. ESBs range in size from a few cm long and a few mm wide to metres in length and >10 cm wide. Boudins are found rarely in the more competent psammitic layers, and are generally symmetrical, with only weak local asymmetry, indicating an approximately top-to-the-SE sense of shear. Folds vary from cm-scale recumbent isoclinal folds to gently inclined open m-scale folds, to upright ptygmatic m-scale folds, although most folds are w0·5 m to 1 m in wavelength, and are tight to isoclinal, gently plunging, and moderately to gently inclined (Fig. 5C). Although there is a large variation in fold hinge azimuth, no sheath folds were found. The foliation is refracted by compositional layering. Deformation styles vary within the shear zone, depending upon the lithology. In pelitic rocks, which are the predominant lithology towards the base of the shear zone, the rocks are highly foliated. In this zone of less competent lithologies, folds are rare and, where seen, are cm-scale and isoclinal (Fig. 5A), consistent with extreme transposition. Towards the top of the shear zone, a more competent psammitic lithology predominates, and larger-scale folds, ESBs and boudins are common. Below the shear zone, where only minor leucosome veins are found in boudin necks, the rocks lack evidence of shear-related deformation. The shear zone can be traced over a strike length of w10 km southeastwards from the hinge of the Schwerin Fold; however it has not been found near the hinge or on the northern limb of the Schwerin Fold. Leucosomes from ESBs within the shear zone have an assemblage comprising quartz–microcline–biotite–muscovite– plagioclase–tourmalinecordierite. Microcline is extremely coarse (>5 mm) and contains w0·2 mm rounded to subidioblastic inclusions of quartz, plagioclase, muscovite and biotite (Fig. 5F). Biotite occurs either as small (w0·5 mm) euhedral ﬂakes or as larger (w3 mm) schlieren, which are unlikely to be crystallisation products, but appear to have been entrained during anatexis and leucosome movement (see Fig. 5E). This biotite is slightly more magnesian (Mg#=50–53) than that from the host rocks (Mg#=33–39; Longridge 2006), and may be the restitic product of the partial melting which formed the leucosomes. Muscovite is commonly poikilitic. Leucosomes

66

LUKE LONGRIDGE ET AL.

MELT SEGREGATION AND MIGRATION IN THE SCHWERIN FOLD

may contain biotite selvedges at their margins. These leucosomes are similar to the small granite sheets found in the migmatite zone in the eastern Bushveld Complex metamorphic aureole, which Harris et al. (2003) attributed to incongruent melting of biotite in the pelites of the Silverton Formation. No post-crystallisation deformation features or any evidence for strain is observed in the leucosomes, indicating that deformation had ceased in the shear zone by the time the leucosomes had crystallised.

4. Structural geology of the Schwerin Fold The broad structure of the Schwerin Fold diﬀers from other comparable structures such as the Katkloof Fold (Fig. 1), in that it has a strongly curved axial trace. At the contact with the RLS, the Schwerin Fold hinge plunges towards the SW, whereas the fold hinge in the core of the fold has a SSE plunge (Fig. 6). Both the eastern and western limbs of the Schwerin Fold give way to open synclines, where the strike of the rocks adjacent to the pericline gently curves into the regional NW–SE strike of the Pretoria Group rocks. A structural investigation by Uken (1998) focused primarily on the non-migmatitic core of the Schwerin Fold, where he noted boudinaged calc–silicate layers and cuspate–lobate folds with radially oriented fold axes. These folds were noted to have an associated axial planar cleavage, and slickenside lineations on bedding surfaces, consistent with a ﬂexural slip fold mechanism (Uken 1998). Also noted was a non-penetrative bedding-parallel S2 cleavage in the limbs of the fold.

4.1. Low-grade metapelites away from the Schwerin Fold In the northeast of the study area, low-grade metapelites are not aﬀected by the Schwerin Fold, and structures are more representative of the regional trends in the ﬂoor rocks to the RLS. In this area, the bedding in metapelites of the Lydenburg Member dips moderately towards the RLS (average orientation 124(/18(SW) and the oblique foliation deﬁned by biotite has an average orientation of 232(/32(SE (Fig. 6B[II]).

4.2. Quartzites adjacent to the RLS contact Adjacent to the RLS contact, competent Magaliesberg Formation quartzites appear to inhibit foliation formation, with the foliation only locally developed and deﬁned by an elongation of quartz grains. Bedding orientations in the hinge of the fold deﬁne a -pole girdle, which gives a fold axis plunge of 47( on an azimuth of 227( (Fig. 6D[II]). Bedding data from the northwestern limb are more scattered than those from the southeastern limb. A best-ﬁt partial -pole girdle through these data deﬁnes a fold plunge of 52( on 346( (Fig. 6D[II]). A progressive change in bedding orientation occurs from w170(/ 78(W in the north to w195(/62(W in the south. Bedding

67

orientations on the northwestern limb are steeper than those on the southeastern limb.

4.3. Metasediments from the core of the Schwerin Fold In the core of the Schwerin Fold, bedding data from the Lydenburg Shale Member show more scatter than data from the quartzites adjacent to the contact. The data deﬁne a -pole girdle with a fold axis plunge of 26( on 188( (Fig. 6A[II]). This is slightly shallower than the hinge-line orientations of m-scale outcrop folds, but the azimuths nonetheless correspond well. Foliations from metapelites of the Lydenburg Member are approximately vertical to steeply W-dipping and N–S trending in the centre of the pericline, but some variation in orientation occurs.

4.4. Shear Zone Within and adjacent to the shear zone to the east of the Schwerin Fold, a number of structural elements are present. Bedding and foliation orientations are variable, but, on average, they are 146(/16(SW and 202(/20(W, respectively (Fig. 6C[II]). Extensional shear band (ESB) orientations are also variable, owing to their listric nature, and give an average orientation of 106(/22(S (Fig. 6C[II]; given the curved character of the ESBs, an average value is likely to be the best representation of their orientation). ESBs are not present as a conjugate set, but display a consistent vergence towards the SSW. Boudin axes plunge shallowly to the east or west. Lineations are also sub-horizontal, and plunge either shallowly NNW or SSE, with the exception of a single outlier, which has an orientation of 30( on 225(. Most lineations are located on bedding surfaces, and are mineral stretching lineations. Fold hinges have highly variable orientations. The plunge of most folds is quite shallow, but azimuths vary from N to W and S.

5. Structural interpretation of the Schwerin Fold In order to understand the orientation of structures in the Schwerin Fold, consideration must be given to the orientation of rocks in the adjacent RLS. A recent re-evaluation of the palaeomagnetism of the Bushveld Complex (Letts 2007), indicates that the RLS acquired its remnant magnetisation (i.e. passed through the Curie temperature) whilst still horizontal. If the fold and shear zone formed before this (see below; also Uken & Watkeys 1997; Uken 1998), then their geometry can only be determined by back-rotating them by an amount equivalent to the average dip of the RLS. Thus, by removing the eﬀects of the regional w20(SW dip of the RLS and its ﬂoor rocks, one can obtain the ‘original’ orientation of the Schwerin Fold.

Figure 4 (A) Hornfels from the lower aureole (below the migmatite zone) showing lithological layering deﬁned by variable biotite content. Spots are chiastolite porphyroblasts. (B) Photomicrograph of (A) showing cordierite and andalusite porphyroblasts, and a weak foliation slightly oblique to bedding (crossed polars; width of ﬁeld of view=4 mm). (C) Folding of migmatitic metapelites in the core of the Schwerin Fold. Note the subvertical orientation of the axial plane to the folds, which mirrors the large-scale structure of the pericline. (D) En-echelon tension gashes oriented oblique to bedding in metapelite from the core of the pericline, in the lower migmatite zone. (E) Enlargement of tension gashes in rocks closer to the contact with the RLS than (D), on the western limb of the fold. (F) Patchy leucosome network no longer restricted to tension gashes (upper migmatite zone). (G) Photomicrograph of metapelite from the pericline core, showing ﬁbrolite seams developed at the expense of biotite (uncrossed polars, width of ﬁeld of view 2 mm). (H) Photomicrograph of metapelite from the core of the pericline, showing a large cordierite porphyroblast with an inclusion-rich core and clear rim, in addition to biotite and corroded quartz, indicating melting (crossed polars; width of ﬁeld of view=2 mm). Mineral abbreviations after Kretz (1983).

68

LUKE LONGRIDGE ET AL.

Figure 5 Outcrop and microscopic structural characteristics of the shear zone: (A) Isoclinal intrafolial folds and transposed bedding in semi-pelitic schist; (B) Oblique earlier foliation in a low-strain lens within quartz-biotite schist. Note coarse quartz grains in the pressure shadow formed by this lens. This texture is interpreted as a composite foliation due to progressive shearing deformation rather than overprinting deformation events (uncrossed polars, width of ﬁeld of view=5 mm); (C) Tight SE-verging folds with a shallow NE-dipping axial planar foliation; (D) Shallow SE-verging extensional shear bands ﬁlled with granitic leucosome; (E) Close-up view of a leucosome within an extensional shear band, containing biotite schlieren which indicate vergence towards the southeast; (F) Photomicrograph of a granitic leucosome, showing a large (w5mm) K-feldspar with euhedral to subhedral quartz, plagioclase and biotite inclusions, indicating crystallisation from a melt (crossed polars; width of ﬁeld of view=2 mm). Mineral abbreviations after Kretz (1983).

5.1. Low grade metapelites away from the Schwerin Fold Metapelites in the contact aureole distal to the Schwerin Fold contain a penetrative subhorizontal cleavage that is typical of the RLS aureole away from periclinal structures (Uken 1998). This fabric is generally subparallel to bedding, but becomes

axial planar to periclinal structures towards the core of these structures, creating a distinctive upward-closing fan pattern (Uken & Watkeys 1997; Fig. 9). Rotation of the structural data through rotation by 20( around a horizontal axis trending 135( results in an

MELT SEGREGATION AND MIGRATION IN THE SCHWERIN FOLD

Figure 6 Geological map of the Schwerin Fold, showing stereonets for the various structural domains. Small stereonets [II]=original data. Large stereonets [I]=original data rotated 20( about a horizontal axis trending 135(. A – Core of the Schwerin anticline; B – Lower-grade schists below migmatite zone; C – Shear zone east of the anticline; D – Fold hinge within the Magaliesberg Formation adjacent to the contact with the RLS. Black circles=bedding; grey diamonds=foliation; white circles=measured fold hinge lines; grey circles=fold hinge lines from -pole girdles; grey horizontal lines=extensional shear bands; grey pentagons=lineations; black squares=boudin long axes. B and C show average orientations for foliation, bedding, and extensional shear bands. The arrow on C[I] indicates the approximate shear direction of the shear zone.

69

70

LUKE LONGRIDGE ET AL.

approximately horizontal bedding dip, and a fabric which dips shallowly away from the Schwerin Fold (Fig. 6B[I]), as expected from the fanning cleavage model (Uken & Watkeys 1997).

5.2. Quartzites adjacent to the RLS contact The scatter of bedding data from the northwestern limb of the fold in the quartzites adjacent to the contact with the RLS may be due to broad curvature of this limb of the fold (Fig. 3), or to smaller, m-scale folding observed rarely on this limb. A m-scale open fold located on the northwestern limb may reﬂect this. It has a hinge orientation of 78( on 337( (Fig. 6D), which is similar to the pole to the best-ﬁt -pole girdle through the bedding data of 52( on 346(. This suggests that the small-scale folding of these quartzites is a consequence of broad-scale warping of the northwestern limb. Rotation of the data does not signiﬁcantly aﬀect the geometry of this domain, but does result in a shallower plunge for the fold axis in this domain.

5.3. Metasediments from the core of the Schwerin Fold Bedding data from the Lydenburg Shale Member show more scatter than data from the quartzites adjacent to the contact with the RLS, because bedding in this domain is rotated by dm- to m-scale folds, and chaotic deformation has occurred, probably reﬂecting melt-assisted strain localisation. Additionally, the variability in the orientation of the ﬁbrolitic foliation is probably due to disruption by melt-assisted strain localisation. On a broad scale, the subvertical fabric present in the core of the Schwerin Fold rotates to the more general shallowlydipping fabric orientation, subparallel to bedding, that is present throughout the lower-grade parts of the RLS aureole (Uken 1998; Fig. 9). It is interesting to note that, whilst the present geometry of the Schwerin Fold is asymmetric, with steeper dips on the northwestern limb and shallower dips on the southeastern limb, rotation of the data to remove postRLS tilting creates a more symmetric fold, with dips on the northwestern limb similar to those on the southeastern limb (Fig. 6A). Such symmetry would be more consistent with diapir development.

5.4. Shear Zone The shallow E or W plunge of boudins from the shear zone (Fig. 6C) is consistent with a broadly N–S extension direction, and this does not change signiﬁcantly following the rotation of the data. Lineation data also remain largely unaﬀected by the rotation, and the single outlier in lineation measurements (30( on 225() coincides with the bedding-cleavage intersection orientation. There is no geographic control on the variability of fold axis orientations, and so plotting a -pole girdle through these data would be meaningless. One possibility is that leucosome veins could have facilitated disruption and shearing-induced rotation of a set of SSE-verging folds (see below). Apart from the folds, the orientation of structures in the shear zone indicates vergence towards the south prior to rotation. This orientation is subparallel to the axial trace in the core of the Schwerin Fold, and does not appear to be consistent with gravity-induced shedding of material oﬀ the ﬂanks of a rising fold, which should be perpendicular to the axis of the fold. The high strains indicate substantial ﬂattening normal to the RLS contact and subhorizontal extension. However, the ESBs and leucosome-ﬁlled shears are not conjugate, but verge asymmetrically to the south and, together with the overall fold asymmetry, indicate a signiﬁcant non-coaxial strain component.

Following back-rotation to correct for the RLS dip, the vergence of the shear zone is towards the southeast (SEdipping ESBs, NW-dipping axial planar foliation – Fig. 6C[I]). This is somewhat more compatible with gravity-induced shedding of material from a rising Schwerin Fold with a SWtrending axis, which is the axial trace of the Schwerin Fold at the contact with the RLS, even following rotation. The sense of shear for this shear zone (verging top-to-the-SE, away from the core of the fold) is the opposite of that expected for typical fold formation via a ﬂexural slip or ﬂexural ﬂow mechanism, where shear sense would be expected to be reverse and top-to-the-NW towards the fold axis. However, diapirism of the ﬂoor rocks and slumping of material from the limb of the rising diapir is consistent with the vergences observed.

6. Structural relationships of the leucosomes The leucosomes within the migmatite zone show abundant evidence of representing crystallised anatectic melts (e.g., Johnson et al. 2003). They are found in a wide variety of structural settings that indicate an intimate timing relationship between deformation and melting. These structural relationships vary according to structural context within the larger Schwerin Fold and shear zone. Within the core of the pericline, leucosomes occur in en-echelon tension gashes, as discordant pods, or as diktyonitic interconnected networks (see also Johnson et al. 2003) between leucocratic bedding layers. Leucosomes intrude sites of localised high strain, where the bedding and foliation are displaced, truncated and rotated (Fig. 7B). Elsewhere, they occur subparallel to the subvertical foliation (Fig. 7A), as well as subparallel to bedding, but they may also be discordant to both bedding and foliation. Leucosome veins that are subparallel to the bedding and the foliation may form a network connecting these two planes. Elsewhere, smaller shear planes with 0·5–1 cm-wide leucosomes displace and rotate bedding. In the shear zone to the southeast of the anticlinal core, leucosome orientations are generally subhorizontal, contrasting with the predominantly subvertical orientation of leucosomes in the core of the Schwerin Fold. Leucosomes are found subparallel to bedding and foliation, and locally pool below more competent psammitic layers (Fig. 7C). They truncate fold hinges along the axial planar fabric, and ﬁll the cores of cm- to dm-scale folds (Fig. 7D). They commonly appear to have intruded preferentially along extensional shear bands (Fig. 5D). Locally, they contain disrupted schlieren of biotite (Fig. 5E) caught up during leucosome movement, that typically display geometries similar to mica ‘ﬁsh’, indicating movement on ESBs with a vergence towards the southeast (SSW prior to rotation). Leucosomes are also locally ptygmatically folded on a cm-scale, and develop in rare boudin necks.

7. Interpretation of the structural relationships of the leucosomes In the core of the Schwein Fold the rocks are dominated by a subvertical axial planar fabric and upright, moderately plunging m-scale folds. Leucosomes occur in en-echelon tension gashes and evidence also exists for melt movement having occurred along the subvertical ﬁbrolite seams (Figs 4D–E, 7B). These features are consistent with melt having been lost upwards along the steep axial planar fabric and bedding in the fold limbs, with only limited entrapment in a few extensional sites. In the shear zone, most structures are much less steep than in the antiform. A number of syn-shearing extensional sites

MELT SEGREGATION AND MIGRATION IN THE SCHWERIN FOLD

71

Figure 7 Leucosome structural sites: (A) Subvertical axial planar cleavage with leucosome from the migmatite zone in the core of the anticline; (B) Leucosomes within shear planes displacing lithological layering in the core of the anticline. Pencil parallel to axial planar foliation deﬁned by ﬁbrolite and aligned andalusite; (C) Leucosomes aligned along the NW dipping foliation in the shear zone, and along bedding; (D) Leucosome-ﬁlled dilational areas in the hinge zone of a recumbent fold in the shear zone.

(fold hinges, ESBs, boudin necks) were exploited by the melt. Additionally, the shear zone is overlain by a thick, competent quartzite layer, which may have acted as a trap for upwardmigrating melts. This, in turn, would have enhanced ductility, concentrating shear strain further and possibly explaining the variable fold orientations as isolated hinges were able to rotate along slip surfaces lubricated by melt.

8. Modelling of the Bushveld Complex thermal aureole Gerya et al. (2003, 2004) numerically modelled diapir development in the ﬂoor rocks beneath the RLS in two dimensions, based on the three-dimensional conceptual model of Uken & Watkeys (1997). This two-dimensional model assumes a 200– 300 kg/m3 density contrast between the dense maﬁc magmas of the RLS and the underlying sedimentary Pretoria Group. The rheological properties and relationships of the model depend on composition, temperature, pressure, strain rate and degree of melting and, hence, resemble the physical properties of the rocks quite well. This is evident in the sensitivity of the model to changes in the initial properties of the rocks (temperature of the ﬂoor rocks and RLS, rheological strength, lateral box size, and amplitude of the initial pericline from which these structures nucleate). A timeframe of w800,000 years was estimated for diapir growth, and a corresponding growth rate of 8 mm/

year for the most well-developed diapirs (e.g. the Phepane Dome; Fig. 1) was calculated based on the amplitude of the diapirs and the maximum timeframe for growth. The results of this modelling are considered to be robust, accurate estimates, which correspond well to the initial estimates of an w6 mm/ year growth rate made by Uken & Watkeys (1997). Since ﬁeld evidence does not indicate any diapir-related structural overprint of the leucosomes in the Schwerin Fold, and there is no evidence that the lower limit of migmatite development has been signiﬁcantly folded by the Schwerin Fold (isograds cut lithological boundaries and are essentially parallel to the RLS contact; Fig. 3), it can be assumed that migmatite development should have occurred over at least a similar timeframe to diapir growth. However, previous thermal modelling for the heating of the Bushveld Complex aureole using a simple one-dimensional conductive cooling approach, where the RLS is intruded as a single sill into the sediments of the Pretoria Group at w3 km depth with a temperature of 1200–1300(C (Johnson et al. 2003; Harris et al. 2003), does not indicate such a lengthy timeframe for migmatite development. The model of Harris et al. (2003), in fact, assumed a RLS sill thickness of only 1500 m, and the model does not simulate temperatures suﬃcient to account for the melting in the aureole if conductive heating occurred. Instead, they suggested extensive ﬂuid circulation could have accelerated the heating rate and, thus, maintained suprasolidus temperatures over signiﬁcant

72

LUKE LONGRIDGE ET AL.

Table 1 Parameters used for the multiple intrusion thermal modelling using Thermos 2·0 ( 1996 Raven Applications). Geothermal Gradient Model Depth Cell Dimension Time Interval Depth of Initial Intrusion Tintrusion Time for Emplacement Thermal Conductivity

30(C/km 20 km 100 m (2000 cells) 1 year 3 km 1160(C to 1300(C 75,000 years 610 3 cal/deg/cm2

distances from the contact. Johnson et al. (2003) assumed a more realistic 7 km thickness for the RLS, and their model simulated temperatures above the solidus at orthogonal distances of >350 m from the contact. However, this model suggested that, at these distances from the contact, suprasolidus temperatures would be maintained for only w350,000 years. Consequently, the heating model of Johnson et al. (2003) would require a growth rate of over 12 mm/year for the Schwerin Fold. This growth rate is double that of the initial estimates made by Uken & Watkeys (1997), and 50% greater than that calculated by Gerya et al. (2003). In response to this divergence between the results from the thermal and rheological modelling the present authors have developed a thermal model of the aureole based on the intrusion of the RLS magmas in multiple pulses, rather than as a single event. This model follows Cawthorn & Walraven (1998) who developed an intrusive–crystallisation sequence based on trace element compositions and isotopic ratios of the RLS, and calculations of the amount of magma required to form the layered chromitite bands in the RLS. Their model suggests that the RLS sill was progressively inﬂated during distinct periods of magma addition interspersed with periods dominated by fractionation (Cawthorn & Walraven 1998).

8.1. Model setup The thermal model was set up using similar parameters to Cawthorn & Walraven (1998) (see Table 1). The modelling was performed using the program Thermos 2·0 ( 1996 Raven Applications). Cell dimensions and time intervals were optimised to give the maximum resolution capable by the program. The total emplacement time is based on the calculations of Cawthorn and Walraven (1998).

8.2. Model results A revised heating history for the RLS aureole has been derived using this setup. These results predict that anatexis would have extended to an orthogonal distance of >500 m from the contact and that, at a distance of 300 m from the contact, anatexis would have persisted until at least 600,000 years after intrusion. Anatexis would have continued for more than 500,000 years at a distance of w400 m from the RLS contact (Fig. 8). This incremental intrusion model predicts higher temperatures over longer timeframes than the thermal eﬀects of a single intrusion, and this increase in the calculated width of the migmatite zone is more compatible with the observed extent of anatexis in the aureole (>500 m, Fig. 3; although the lower limit of anatexis does vary somewhat along strike). More signiﬁcantly, however, the calculated timeframe for melt generation more closely approximates the rheological modelling results for diapiric fold formation obtained by Gerya et al. (2003, 2004).

9. Discussion Deformation is well known as a key factor in creating mechanisms and pathways by which buoyant anatectic melt can segregate from its source rocks and escape upwards in orogenic terrains (e.g. Brown 1994; Kisters et al. 1998). The Schwerin Fold region shows that similar relationships can also apply in a contact metamorphic environment. The leucosome geometries in the migmatites from the Schwerin Fold core and the area to the southeast suggest that highly variable local stresses were an important controlling factor in the movement and emplacement of granitic melts beneath the Bushveld Complex intrusion. Both in the anticlinal core of the Schwerin Fold, as well as in the shear zone, leucosome distribution is structurally controlled. In the anticlinal core, en-echelon tension gashes provided low pressure sites where leucosome accumulated. However, the subvertical axial planar foliation and upright folds characteristic of the core of anticline provided pathways for leucosome to migrate upwards. The limited preservation of leucosomes in these features (apart from the tension gashes, only the melt escape pathways are preserved as ﬁbrolite seams) suggests very eﬀective melt migration once it entered these structures. This is perhaps not surprising, as the thermal gradient in the aureole is inverted, so that melts migrating upwards were actually entering hotter rocks. In the shear zone to the east of the Schwerin Fold, leucosomes are clearly located in dilational structural features such as boudin necks and ESBs that formed during subhorizontal shearing. The generally subhorizontal nature of the lithological layering and structural fabrics in this area has resulted in pooling of leucosome below more competent lithologies, with only limited upwards migration possible along the shallowly-dipping axial planar foliation in the shear zone. This concentration of leucosome in the shear zone would have increased ductility, thus concentrating further deformation in the shear zone. It is clear from Figure 9 that the stress pattern in and around the Schwerin Fold during Bushveld metamorphism was highly variable and cannot be readily explained by a simple NW–SE compression mechanism as proposed by, e.g., Sharpe & Chadwick (1982). The leucosomes provide evidence that subvertical and subhorizontal foliations were developing simultaneously in diﬀerent parts of the structure, and the foliation in the subsolidus part of the aureole displays similar syn-peak timing to the other structures but has a signiﬁcantly diﬀerent orientation (shallow dip obliquely away from the RLS contact). Uken & Watkeys (1997) noted similar complex fabric relationships in and around the Katkloof Fold (Fig. 1), with a subvertical axial planar fabric in the fold core rotating outwards to shallower dips until, in the interdomal areas, it is approximately bedding-parallel. These interdomal areas contain extensional features, such as boudins and conjugate ESBs, that are indicative of predominantly vertical compression and horizontal extension via pure shear (Uken 1998). In contrast, the core of the Katkloof Fold contains a variety of constrictional fold and mullion structures in addition to the subvertical foliation. Such observations suggest that fold development was driven by buoyancy, rather than regional compression. Uken & Watkeys (1997) proposed that the diapiric folds nucleated in the necks between NE-trending ﬁnger-like intrusions of the RLS. More recently, Clarke et al. (2009) have demonstrated the link between aureole structures and intrusion geometry on a smaller scale around Burgersfort (Fig. 1). They attributed the strain heterogeneity in the aureole rocks to the twin causes of the stepped nature of the intrusive contact, which locally cuts across the Pretoria Group stratigraphy, and subsequent magmatic inﬂation of the individual ﬁngers of

MELT SEGREGATION AND MIGRATION IN THE SCHWERIN FOLD

73

Figure 8 Results of thermal modelling of the RLS aureole, using the single intrusion model of Johnson et al. (2003) (grey lines), and the multiple intrusion model of Cawthorn & Walraven (1998) (black lines). Note that for the single intrusion model peak temperatures are higher and are achieved earlier, but cool below the solidus quicker (<400,000 years after the intrusion of the RLS). In the case of the multiple intrusion model the solidus is exceeded further from the contact and, at 300 m from the contact, suprasolidus temperatures are maintained for w600,000 years. Irregularities in the curves are due to the periodic emplacement of magma into the RLS chamber, as described in Cawthorn & Walraven (1998).

the RLS sheet intrusion. Where the intrusion ﬁngers were separated by country rock, or where a step occurred, lateral inﬂation induced compressional stresses in the country rock, generating symmetric or asymmetric structures depending on the exact intrusion geometry. A local stress regime induced by the geometry of the intrusion can more readily explain such features as the curvature of the Schwerin Fold axial trace and the restricted extent of the shear zone orthogonal to the Schwerin Fold axis. It could also explain the subsolidus foliation in the northeastern part of the study area, which dips obliquely away from, rather than towards, the main RLS contact. If Clarke et al’s (2009) model is applied, this foliation was induced by a (now-eroded) transgressive RLS contact dipping southeast and lying southeast of the study area. The exact timing of development of this lower-grade foliation relative to the axial planar and shear foliations is a matter of debate. Figure 8 shows that the metamorphic peak is attained progressively later in rocks further from the RLS contact; however, it is likely that peak conditions were reached before termination of deformation related to the formation of the fold. The fact that the andalusite, cordierite and garnet porphyroblasts in the lower-grade rocks show a late- to postkinematic timing relative to this foliation (compared with the pre- to syn-kinematic relationships in both the fold and the shear zone), may reﬂect the delayed thermal response in the lower aureole compared with the rapid transfer of stresses that

would ultimately have driven foliation development once appropriate metamorphic conditions were reached. The Schwerin Fold is, thus far, the only periclinal structure in the Bushveld Complex aureole that appears to be associated with a large shear zone. As shown in the diagrammatic representation of the structures in the Schwerin Fold (Fig. 9), the development of the shear zone to the east of the pericline is compatible with slumping of ductile material from the ﬂanks of a rising diapir. This is supported by the restricted areal distribution of the shear zone, which disappears towards the southeast away from the fold. Similar slumping eﬀects have been noted by Clarke et al. (2005) adjacent to the Steelpoort Pericline, but only in magmatic RLS rocks. The western limb of the Schwerin Fold appears to be extremely attenuated (Fig. 3) but contains no evidence of shearing, although outcrop of non-quartzitic rocks is poor. If the Schwerin shear zone is unique, the reason may lie in the proximity of the thick quartzite unit of the Magaliesberg Formation to the RLS contact in the Schwerin area. This unit may have acted both as a barrier to rising anatectic melts and as a coherent structural unit which then slid southeastwards oﬀ the diapirically rising Schwerin pericline, thus focusing shear stresses in the meltweakened unit immediately below it. In other periclines, the RLS contact cuts across more-ﬁnely-bedded, less refractory, lithologies which would have undergone general melt-induced weakening that, together with their original rheological

74

LUKE LONGRIDGE ET AL.

Figure 9 Diagrammatic summary of the structural features found within the study area: (A) The NE domain furthest from the RLS contains a weak ESE-dipping foliation and subhorizontal bedding; (B) The core of the anticline contains dm- to m-scale folds and a subvertical axial planar cleavage; (C) The shear zone contains SE-verging, tight to isoclinal folds with an axial planar cleavage. Bedding-cleavage intersection lineations and movement (mineral extension) lineations are present on bedding surfaces. ESBs have a SE vergence. Bedding is subhorizontal (following restoration by rotation – see Fig. 6); (D) Schematic representation of the Schwerin Fold showing subvertical leucosomes oriented axial planar to folds in the core of the pericline, with the top-to-the-SE shear zone resulting from gravitational slumping of material oﬀ the eastern limb of the fold, and RLS layering truncated and folded by the diapiric fold.

properties, would have distributed the limb shear stresses more homogeneously. Another question which arises from the Schwerin Fold is whether the leucosome present in the migmatites contributed to the overall buoyancy of the fold, and accelerated the diapiric rise of the ﬂoor rocks. Johnson et al. (2003) noted that Marginal Zone norites of the RLS are brecciated and invaded by signiﬁcant quantities of leucosome near the Derde Gelid

Pericline south of the study area (Fig. 1). A similar concentration of leucosomes is not seen in the quartzites or RLS in the hinge of the Schwerin Fold, but the eﬃciency of melt segregation and migration attested to by the ﬁbrolite seams in the core of the fold makes it most likely that such an accumulation did form in the now-eroded crest of the fold. Notwithstanding this, when a buoyant melt is created and segregated from its source rock, a denser residue/restite is

MELT SEGREGATION AND MIGRATION IN THE SCHWERIN FOLD

left behind. In this case, the overall density of the rock volume does not change, as the buoyancy of the magma is counteracted by the increased density of the restite. Field relations indicate that the Schwerin Fold encompasses the entire volume of suprasolidus rock and, thus, that no signiﬁcant buoyancy eﬀect is likely to have existed due to melt extraction. Furthermore, Johnson et al. (2003) estimated that overall melting in the Schwerin Fold is unlikely to have exceeded 10 vol%. The primary impetus for diapirism must, thus, have been the density contrast (200–300 kg/m3; Gerya et al. 2003) between the upper Pretoria Group ﬂoor rocks and the overlying RLS, with an irregular lower RLS contact created by the ﬁnger-like geometry of the magma intrusions (Uken & Watkeys 1997; Clarke et al. 2009). The interrelationship between the thermal and deformation eﬀects in the upper contact aureole caused by the intrusion of the RLS must be evaluated within the context of a model that acknowledges the multi-stage nature of inﬂation of the Bushveld magma chamber (Cawthorn & Walraven 1998). In contrast to the single-stage intrusion models (e.g., Johnson et al. 2003; Harris et al. 2003), which would require doming rates double that calculated by Gerya et al. (2003), the multiple-intrusion model generates slightly lower peak temperatures but a longer-lived metamorphic eﬀect that achieves very good agreement with the timescale for diapiric fold development deduced by Gerya et al. (2003).

10. Conclusions The segregation and migration of melt in the contact aureole beneath the RLS is intimately linked to the structural development of the aureole, particularly in areas such as the Schwerin Fold, where structures are more pronounced, and where deformation is clearly contemporaneous with metamorphism and migmatisation within the aureole. Of additional importance is the absolute timing of metamorphism, migmatisation and diapir development in the contact aureole. The Schwerin Fold developed over w600 000 years, with a growth rate of w8 mm/year, which is consistent with modelled estimates for the duration of diapir development elsewhere in the aureole (Gerya et al. 2003). A simple, one-dimensional, cooling model of a 7–8 km-thick sill does not account for this – rather, a model of multiple intrusive pulses for the RLS is more consistent with the longer duration of diapir development. It is evident that migration and accumulation of buoyant leucosome is controlled by the structural geometry of the area in which migmatisation is occurring – subvertical structures allow these leucosomes to escape upwards, whilst subhorizontal structures result in the accumulation of leucosome, in turn triggering a local increase in ductility. The latter may well have led to slumping along the diapir ﬂank being localised in a broad extensional shear zone.

11. Acknowledgements We are grateful to Anglo Platinum for logistical support during ﬁeld work for this study. Tim Johnson and Chris Gerbi are thanked for constructive reviews which improved the quality of the manuscript.

12. References Bai, Q., Jin, Z.-M. & Green II, W. 1997. Experimental investigation of the rheology of partially molten peridotite at upper mantle pressures and temperatures. In Holness, M. B. (ed.) Deformation-

75

enhanced Fluid Transport in the Earth’s Crust and Mantle, 40–61. The Mineralogical Society Series 8. London: Chapman & Hall. Bhadra, S., Das, S. & Bhattacharya, A. 2007. Shear Zone-hosted Migmatites (Eastern India): the Role of Dynamic Melting in the Generation of REE-depleted Felsic Melts, and Implications for Disequilibrium Melting. Journal of Petrology 48, 435–57. Brown, M. 1994. The generation, segregation, ascent and emplacement of granite magma: the migmatite-to-crustally-derived granite connection in thickened orogens. Earth Science Reviews 36, 83–130. Brown, M. 2005. Synergistic eﬀects of melting and deformation: an example from the Variscan belt, western France. In Gapais, D., Brun, J. P. & Cobbold, P. R. (eds) Deformation Mechanisms, Rheology and Tectonics: from minerals to the Lithosphere, 205–26. Geological Society, London, Special Publication 243. Bath, UK: The Geological Society Publishing House. Brown, M., Averkin, Y. A., McLellan, E. L. & Sawyer, E. W. 1995. Melt segregation in migmatites. Journal of Geophysical Research 100, 15655–79. Brown, M. & Solar, G. S. 1998. Shear zone systems and melts: feedback relations and self-organization in orogenic belts. Journal of Structural Geology 20, 211–27. Button, A. 1978. Diapiric structures in the Bushveld, north-eastern Transvaal. Information Circular, Economic Geology Research Unit 96. Johannesburg, South Africa: University of the Witwatersrand. 13 pp. Cawthorn, R. G. & Walraven, F. 1998. Emplacement and crystallization time for the Bushveld Complex. Journal of Petrology 39, 1669–87. Clarke, B. M., Uken, R., Watkeys, M. K. & Reinhardt, J. 2005. Folding of the Rustenburg Layered Suite adjacent to the Steelpoort pericline: implications for syn-Bushveld tectonism in the eastern Bushveld Complex. South African Journal of Geology 108, 397–412. Clarke, B. M., Uken, R. & Reinhardt, J. 2009. The geometry and emplacement mechanics of a Bushveld Complex peridotite body. South African Journal of Geology 112, 141–62. Eriksson, P. G., Schweitzer, J. K., Bosch, P. J. A., Schereiber, U. M., Van Deventer, J. L. & Hatton, C. J. 1993. The Transvaal Sequence: an overview. Journal of African Earth Sciences 16, 25–51. Gerya, T. V., Uken, R., Reinhardt, J., Watkeys, M. K., Maresch, W. V. & Clarke, B. M. 2003. Cold ﬁngers in hot magma: numerical modeling of country rock diapirs in the Bushveld Complex, South Africa. Geology 31, 753–6. Gerya, T. V., Uken, R., Reinhardt, J., Watkeys, M. K., Maresch, W. V. & Clarke, B. M. 2004. ‘Cold’ diapirs triggered by intrusion of the Bushveld Complex: Insight from two-dimensional numerical modeling. In Whitney, D. L., Teyssier, C. & Siddoway, C. S. (eds) Gneiss Domes in Orogeny, 117–27. Geological Society of America Special Paper 380. Boulder, Colorado & Lawrence, Kansas: Geological Society of America & University of Kansas Press. Gleason, G., Bruce, V. & Green, H. W. 1999. Experimental investigation of melt topology in partially molten quartzofeldspathic aggregates under hydrostatic and non-hydrostatic stress. Journal of Metamorphic Geology 17, 705–22. Hand, M. & Dirks, P. H. G. M. 1992. The inﬂuence of deformation on the formation of axial-planar leucosomes and the segregation of small melt bodies within the migmatitic Napperby Gneiss, central Australia. Journal of Structural Geology 14, 591–604. Harris, N., McMillan, A., Holness, M., Uken, R., Watkeys, M., Rogers, N. & Fallick, A. 2003. Melt generation and ﬂuid ﬂow in the thermal aureole of the Bushveld Complex. Journal of Petrology 44, 1031–54. Hartzer, F. J. 1995. Transvaal Supergroup inliers: geology, tectonic development and relationship with the Bushveld Complex, South Africa. Journal of African Earth Sciences 21, 521–47. Holyoke, C. W. & Rushmer, T. 2002. An experimental study of grain scale melt segregation mechanisms in two common crustal rock types. Journal of Metamorphic Geology 20, 493–512. Johnson, T. E., Gibson, R. L., Brown, M., Buick, I. S. & Cartwright, I. 2003. Partial Melting of metapelitic rocks beneath the Bushveld Complex, South Africa. Journal of Petrology 44, 789–813. Johnson, T., Brown, M., Gibson, R. & Wing, B. 2004. Spinelcordierite symplectites replacing andalusite: evidence for meltassisted diapirism in the Bushveld Complex, South Africa. Journal of Metamorphic Geology 22, 529–45. Kisters, A. F. M., Gibson, R. L., Charlesworth, E. G. & Anhaeusser, C. R. 1998. The role of strain localization in the segregation and ascent of anatectic melts, Namaqualand, South Africa. Journal of Structural Geology 20, 229–42.

76

LUKE LONGRIDGE ET AL.

Kretz, R. 1983. Symbols for rock-forming minerals. American Mineralogist 68, 277–79. Longridge, L. 2006. Structural and metamorphic evolution of the Schwerin Fold, northeastern Bushveld Complex, Limpopo Province. Honours Thesis, University of the Witwatersrand, Johannesburg, South Africa. 42p˙. Letts, S. A. 2007. The palaeomagnetic significance of the Bushveld Complex and related 2 Ga magmatic rocks in ancient continental entities. PhD Thesis, University of the Witwatersrand, Johannesburg, South Africa. 228 pp. du Plessis, C. P. & Walraven, F. 1990. The tectonic setting of the Bushveld Complex in southern Africa, part 1. Structural deformation and distribution. Tectonophysics 179, 305–19. Rosenberg, C. L. & Berger, A. 2000. Syntectonic Melt Pathways in Granitic Gneisses, and Melt-Induced Transitions in Deformation Mechanisms. Physics and Chemistry of the Earth 26, 287–93. Sawyer, E. W. 1991. Disequilibrium melting and the rate of meltresiduum separation during migmatization of maﬁc rocks from the Grenville Front, Quebec. Journal of Petrology 32, 701–38. Schwellnus, J. S. I., Engelbrecht, L. N. J., Coertze, F. J., Russell, H. D., Malherbe, S. J., van Rooyen, D. P. & Cooke, R. 1962. The geology of the Oliphants River area. Explanation to Sheets 2429B

(Chuniespoort) and 2430A (Wolkberg). Geological Survey of South Africa. Sharpe, M. R. & Chadwick, B. 1982. Structures in Transvaal Sequence rocks within and adjacent to the eastern Bushveld Complex. Transactions of the Geological Society of South Africa 85, 29–41. Uken, R. 1998. The Geology and Structure of the Bushveld Complex metamorphic aureole in the Olifants River area. PhD Thesis, University of Natal, Durban, South Africa. 277 pp. Uken, R. & Watkeys, M. K. 1997. Diapirism initiated by the Bushveld Complex, South Africa. Geology 25 723–6. Vernon, R. H., Collins, W. J. & Richards, S. W. 2003. Contrasting magmas in metapelitic and metapsammitic migmatites in the Cooma Complex, Australia. Visual Geosciences 8, 1–22. Walraven, F., Armstrong, R. A. & Kruger, F. J. 1990. A chronostratigraphic framework for the north-central Kaapvaal craton, the Bushveld Complex and Vredefort structure. Tectonophysics 171, 23–48. Waters, D. J. & Lovegrove, D. P. 2002. Assessing the extent of disequilibrium and overstepping of prograde metamorphic reactions in metapelites from the Bushveld Complex aureole, South Africa. Journal of Metamorphic Geology 20, 135–49.

MS received 12 December 2007. Accepted for publication 10 June 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 77–103, 2010 (for 2009)

Parental magmas of Grenville Province massif-type anorthosites, and conjectures about why massif anorthosites are restricted to the Proterozoic Jean H. Be´dard Geological Survey of Canada, CGC-Que´bec, 490 de la Couronne, Que´bec, Canada, G1K 9A9 Email: [email protected] ABSTRACT: Trace element inversion modelling of Grenvillean anorthosite massifs and associated rocks yield NMORB-normalised trace element proﬁles enriched in highly incompatible elements; commonly with negative Nb and Th anomalies. Model melts can be divided into subtypes that cannot be linked through fractional crystallisation processes. Most model melts are depleted in the heavy rare-earth elements and can be explained by partial melting of arc basaltic sources (5–60 melting %) with garnet-bearing residues. Some of the model melts have ﬂat NMORB-normalised proﬁles (for rare-earth elements), have high compatible element contents, and might have been derived from mantle fertilised by arc magmatism, followed by low-pressure fractional crystallisation. Intermediate Ce/Yb types may represent mixtures of these end-members, or less probably, variations in the crustal source composition and residual assemblage. The active tectonic context now favoured for the Grenville Province appears to be inconsistent with plume or thermal insulation models. The heat source for crustal and mantle melting could record either post-orogenic thermal relaxation of a tectonically-thickened arc crust, or basaltic underplating caused by delamination of a mantle root or subduction slab beneath this arc crust. In this context, pre-Proterozoic anorthosites may be lacking, because prior to ca. 2·5 Ga, the crust may have been too weak to be thickened tectonically. The absence of post-Proterozoic anorthosites may be due to the secular decrease in radiogenic heating and cooling of the mantle and crust. KEY WORDS: elements

apatite, crust, gabbro, granite, ilmenite, immiscibility, melting, tectonics, trace

The Proterozoic Eon is characterised by voluminous anorthositic massifs (Fig. 1; An30–70), that are associated with mangerite and granite batholiths, with bodies of gabbro, troctolite and norite, and with minor Fe-rich ferrodiorite to monzonorite (e.g. Morse 1975, 1982; Emslie 1978, 1980, 1985; Wiebe 1992; Ashwal 1993). These rocks can be grouped together as the AMCG suite (anorthosite, mangerite, charnockite, granite or gabbro: Emslie 1985). The petrogenesis of this suite and reason for the restricted age spectrum of massif-type anorthosites (2·5–0·9 Ga: see review in Ashwal 1993 and in Scoates & Chamberlain 1997) are enduring problems (Bowen 1917). In its simplest form, the problem can be summarised as: if anorthosites form by a conventional petrogenetic mechanism (basaltic underplating, plumes, crustal anatexis . . .), then why are they restricted to the Proterozoic? Historically, mantle-derived tholeiitic or high-Al basalt parents for the anorthosites were favoured (e.g. Morse 1969, 1982; Emslie 1980, 1985; Emslie et al. 1994; Wiebe 1990a, b; Olson 1992; Mitchell et al. 1995, 1996; Icenhower et al. 1998). However, a growing minority favour crustal sources (Michot 1965; Berg 1969; Philpotts 1969; Frith & Currie 1976; Simmons & Hanson 1978; Taylor et al. 1984; Demaiﬀe et al. 1986; Duchesne et al. 1989; Scha¨rer et al. 1996; van der Auwera et al. 1998; Longhi et al. 1999; Duchesne et al. 1999; Peck & Valley 2000; Schiellerup et al. 2000; Selbekk et al. 2000; Be´dard 2001; Peck et al. 2004).

2009 Her Majesty The Queen in right of Canada.

To account for the extensive volumes of feldspar, those favouring a crustal origin consider that the source is either inherently aluminous (e.g. feldspar-rich meta-cumulates: Berg 1969; Duchesne et al. 1999; Selbekk et al. 2000), or becomes so as a result of prior anatexis (e.g. Frith & Currie 1976). Among advocates of a mantle origin, many suggest that the feldspar was concentrated as a ﬂotation cumulate during extensive fractionation of aluminous basalt in the lowermost crust (e.g. Emslie 1980, 1985; Duchesne 1984; Longhi et al. 1993; Ashwal 1993; Scoates & Frost 1996). Another way of helping plagioclase to form in abundance is through combined fractionation–assimilation of aluminous crustal rocks (Bowen 1922; Buddington 1936; Michot 1965; Philpotts 1969; Emslie & Hegner 1993; Emslie et al. 1994; Brandriss & Cawthorn 1996; Dempster et al. 1999; Li et al. 2000). Once formed, this concentration of plagioclase could be remobilised and ascend as a crystal-rich mush (Martignole & Schrijver 1970; Woussen et al. 1981). In contrast, others have proposed that the accumulation process is not mechanical, but chemical, a consequence of delayed nucleation (Morse 1982) or prior resorption of feldspar (Wiebe 1988, 1990b; Markl & Frost 1999), with feldspar then being produced in great abundance when it does begin to crystallise upon ascent and emplacement. The lack of superimposed deformation in some anorthosite provinces led to an early consensus for an anorogenic setting (Bridgwater & Windley 1973; Berg 1977; Emslie 1978, 1985;

doi:10.1017/S1755691009016016

78

JEAN H. BEDARD

Figure 1 Location map, adapted from the Grenville map of Tony Davidson (1998), complemented with other sources (Tom Clark, MRNFQ; David Corrigan, GSC, pers. comm.; and a variety of sources quoted in the reference list). SIGE u OM samples are the small dots. Other data locations and source of age data are in the original papers (see text). WB=White Bear Arm anorthosite (Kamo et al. 1996), LF=Lac Fournier lobe, RR=rivie`re Romaine lobe.

Morse 1982; Olson 1992; Scha¨rer et al. 1996; Amelin et al. 1997). However, convincing evidence of rifting is typically absent (Emslie 1978), leading to suggestions of an origin from mantle plumes (Morse et al. 1988) or mantle superswells produced by thermal insulation under a supercontinent (Anderson 1983; Hofmann 1989; Vigneresse 2005). Recent work challenges the anorogenic consensus; suggesting that most anorthosites are late- to post-orogenic (Martignole 1996); related to crustal thickening and delamination (Scoates & Chamberlain 1997; Corrigan & Hanmer 1997), and/or to crustal imbrication (Duchesne et al. 1999). Perhaps the most serious obstacle to progress on all aspects of anorthosite petrogenesis is the lack of liquid compositions amenable to quantitative modelling and testing. This paper corrects this lacuna by calculating the trace element contents of the melts from which Grenville Province anorthosite suite rocks crystallised; complementing a previous study of Nain Plutonic suite rocks (Be´dard 2001). The anorthositic massifs of the Grenville Province are the largest and most voluminous on Earth, and epitomise the anorthosite problem. With such model liquid data in hand, quantitative tests are designed to ascertain whether or not diﬀerent facies are related by fractionation, and melting models are constructed to test whether mantle or crustal sources are most plausible.

1. The Grenville Province and its anorthosites The emerging picture for the tectonic evolution of the Proterozoic Grenville Province is of an active history of cratonward subduction, punctuated by periodic back-arc rifting and accretion of oceanic arc terranes, and terminated by continent– continent collision (e.g. Martignole 1996; Rivers 1997; Rivers

& Corrigan 2000; Hanmer et al. 2000; Culshaw et al. 2000; Ketchum et al. 2002; Gower & Krogh 2002; Dickin & McNutt 2007; Corriveau et al. 2007; Schulz & Cannon 2007; Davidson 2008). When considered in this context, the widespread Grenville Province AMCG complexes appear to record: (a) distal extensional events linked to an active margin setting; (b) back-arc rifting; or (c) are related to post-orogenic collapse of the collisional orogen and delamination of its lithospheric root (e.g. Corrigan & Hanmer 1997; McLelland et al. 1996; Rivers & Corrigan 2000; Gower & Krogh 2002). Ages of AMCG massifs modelled herein range from ca. 1·62 Ga Post-Labradorian (Mealy Mountains complex: James et al. 2000), to ca. 1·36–1·33 Ga Post-Pinwarian (Rivie`re Pentecoˆte, De La Blache: Nantel & Martignole 1991; Gobeil et al. 2002), to ca. 1·18–1·13 Ga Post-Shawinigan (Vanel, Lac St. Jean, Havre St. Pierre: van Breemen & Higgins 1993; Higgins et al. 2002; He´bert & van Breemen 2004a, b), and ca. 1·02–1·008 Post-Ottawan (Mattawa, Labrieville: Owens et al. 1994; He´bert et al. 2005); and thus spans nearly the entire age-range of Grenville Province AMCG magmatism. The Baie-Comeau or Vallant anorthosite massif is undated, and could belong to any of the post-1·36 Ga series. Individual anorthosite massifs in the Grenville can be over 100 km in diameter, and the giant Lac St. Jean complex underlies 20 000 km2 (Fig. 1). The anorthosite massifs are spatially associated with granitic rocks, commonly pyroxene-bearing high-temperature types, which locally form sheaths around the anorthosites (Fig. 1; e.g. Emslie 1991; Emslie & Hunt 1990). Maﬁc cumulate rocks are also commonly associated spatially and temporally with the anorthosites (Fig. 1). These range from various types of leuco-gabbroic rocks, to mela-gabbros, pyroxenites and peridotites, which are commonly rather

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

Fe-rich (e.g. Olson & Morse 1990; Francis et al. 2000). Concentrations of Fe–Ti-oxide minerals, some economic, are also present in and around the anorthositic massifs (e.g. Owens & Dymek 1992; Dymek & Owens 2001; He´bert et al. 2005; Corriveau et al. 2007); and some are associated with sulphide mineralisation (Corriveau et al. 2007). Altogether, rocks of the AMCG suite occupy more than 20% of the Grenville Province (Martignole 1996); and any geotectonic model must account for their presence.

2. Methodology 2.1. The equilibrium distribution method (EDM) The chemistry of many plutonic rocks reﬂects the nature and proportions of accumulated minerals when the proportion of melt entrapped in the cumulate framework is small. Be´dard (1994, 2001) developed the equilibrium distribution method (EDM) to calculate the composition of trace elements in the liquid from which cumulates formed. Its application requires: a set of trace element crystal/liquid partition coeﬃcients (Table 1); a whole-rock trace element analysis (Table 2; for data see Appendix Table A1 – Supplementary Material); a mode (Table 2; Table A1); and an assumed value for the trapped melt fraction (TMF). Uncertainty with regard to the TMF has a signiﬁcant eﬀect on absolute abundances of trace element contents, but has little eﬀect on trace element ratios (Fig. 2b). A TMF value of 0 was assumed for analysed plagioclase megacrysts (e.g. Owens & Dymek 2001). A judicious choice of partition coeﬃcients is paramount. It is now understood that D values vary with melt and mineral chemistry, and it is of critical importance that the D values should be appropriate to the rock being modelled. Each model in this paper was calculated with its own D dataset pinned by the plagioclase An-content. An example for An50 plagioclase is given as Table 1. Since most samples being modelled were not analysed for their mineral chemistry, the An-content of the plagioclase (needed to calculate Ds for plagioclase using the equations in Be´dard (2006a), Table 1) was estimated from norm calculations. Where mineral–chemical data exist, only small diﬀerences from the normative values were observed, validating the use of normative plagioclase compositions for this purpose. The impact of uncertainties in the An content on computed values of D is considered in Figure 2a, and is small. Aside from analytical uncertainty, the largest source of uncertainty in the EDM models computed here is uncertainty in the D regressions vs. An-content (Fig. 2c), which nonetheless is small enough that the main conclusions of this paper are not aﬀected. To calculate values of D for olivine (Be´dard 2005) and orthopyroxene (Be´dard 2007), it is necessary to estimate melt SiO2 and MgO contents. Since most analyses are of cumulate rocks that do not represent frozen liquids, melt compositions must somehow be estimated. Francis et al. (2000) analysed a set of gabbroic chilled margins and dykes that they inferred were related to AMCG magmatism. Normative plagioclase compositions from these were regressed against melt MgO, allowing melt MgO content to be calculated as follows: MgO (wt%) of melt=An (molar fraction) 14·066 (3·951)+0·3663 (0·8966) (1) with an R2 of 0·6788 for N=8. The trend of melt SiO2 vs. An content is not systematic for this dataset and the Francis et al. (2000) dyke data yield a very low average SiO2 of 45·6 wt%. As will become apparent, a crustal derivation is inferred for many melts, and so a more evolved melt SiO2 of 55 wt% was used for all calculations where SiO2 was needed. These are only needed for DZn in Cpx and for apatite Ds. Some of the D values for

79

clinopyroxene (MS in preparation) and orthopyroxene require that the Aliv content be known. The average Aliv contents of clinopyroxene and orthopyroxene from the Labrieville massif analysed by Owens & Dymek (2001) are 0·02 and 0·05, respectively, and these values were used throughout. More precise models would result if mineral–chemical data were available, but the uncertainty in Opx Aliv content only aﬀects D values for Nb and Cu (Table 1). On the other hand, many Cpx D values are controlled by Cpx Aliv content (Table 1). However, since Cpx is rarely present as a model residual phase, few models are aﬀected by this uncertainty. D values for magnetite, ilmenite, apatite and zircon are either constants taken from the compilation of Be´dard (2006b); or are newly derived (unpublished data, >‘T2# in Table 1). To this end, mineral partitioning data were compiled from the literature and values of LnD were regressed against MgO or SiO2 in the liquid; or where no regression was possible, simple averages were computed. The DREE–Y (REE=rare earth elements) data were smoothed using the Lattice Strain Model of Blundy & Wood (1994; for method see Be´dard 2007), and values of these constants will be published elsewhere. Since thin sections and other petrographic details were not generally available, and grain sizes are commonly so coarse as to make thin sections unrepresentative, modes were estimated on the basis of CIPW norm calculations, adjusted for minor elements (see Be´dard 2001 for the method). Speciﬁcally, small proportions of orthoclase were added to plagioclase, pyroxene modes were redistributed to compensate for exsolution, and small amounts of normative chromite, ilmenite and magnetite were distributed among the dominant ferromagnesian silicates. Where larger proportions of magnetite, ilmenite or apatite are present in the norm, a cumulus origin was inferred and integrated into the inverse models. Where reported in the original study, modal data were used to guide reconstruction of the mode. Note that small changes in geochemistry can lead to large variations in normative olivine/orthopyroxene ratios. Furthermore, peritectic and subsolidus reactions may change the orthopyroxene/olivine ratio. However, variations of the olivine/ orthopyroxene ratio have only a minor impact on the computed melt trace element distribution because of the general similarity of orthopyroxene and olivine D proﬁles and values. The basis of the EDM (Be´dard 1994) is a mass balance equation that reconstructs the composition of the melt from which the cumulus minerals were derived, at the point where this assemblage of cumulates+trapped melt was sealed oﬀ from the main body of magma and became a closed system. It is assumed that equilibrium prevailed; that most of the intercumulus melt was quickly expelled from the cumulus framework at near-liquidus temperatures (before it could diﬀerentiate signiﬁcantly); that melt entrapped in the pores of the cumulate framework crystallised in situ; and that there are no complicating post-cumulus percolation or metasomatic eﬀects. If these assumptions hold true, then the EDM yields the trace element composition of the liquidus melt. The pertinent equations and the methodology are explained in Be´dard (1994, 2001), and a revised set of equations that consider the structure-forming role of TiO2 in ilmenite, P2O5 in apatite and ZrO2 in zircon are presented in Be´dard et al. (2009). To preserve mass, a proportion of minerals equivalent to the trapped melt fraction (TMF) needs to be subtracted from the solid assemblage. This backstripping of the solid assemblage is conceptually identical to non-modal melting, and leaves a residual mode complementary to the model melt. As backstripping progresses solid phases may disappear, and so the melting mode must change accordingly. The melting modes used here are either experimentally-determined cotectic proportions, or approximations derived from phase proportions in common

DCs DK DRb DBa DTh DU DNb DTa DLa DCe DPr DPb DSr DP

0·00046 0·0116 0·00046 0·0085 0·0112 0·0054 0·0080 0·0141 0·0853 0·136 0·202 0·290 0·0138 0·0411

Aliv MgO Aliv Av. MgO Av. MgO MgO L, Aliv L, Aliv L, Aliv Av. Av. Av.

Cpx

0·0918 0·242 0·0602 0·540 0·104 0·0755 0·0941 0·0184 0·125 0·121 0·114 0·657 3·06 0·107

16, An 15c, An 17, An 18a, An 37, An 38a, An 32a, An 33, MSAn 64, L, An 64, L, An 64, L, An 20, An 19a, An 39, An

Plag 0·0110 0·0100 0·0100 0·0100 0·00334 0·0035 0·00691 0·00686 0·00224 0·00386 0·00645 0·0570 0·00331 0·0252

63, MgO set=DBa set=DBa 65, MgO 18, MgO 21, MgO 28, Aliv 29, MgO 135–7, L, MgO 135–7, L, MgO 135–7, L, MgO 97, MgO 60, MgO 23, MgO

Opx 0·00236 0·00236 0·00236 0·00097 0·00962 0·0139 0·00181 0·0497 0·00006 0·00014 0·00032 0·00336 0·00404 0·00550

MgO MgO MgO MgO MgO MgO MgO MgO L, MgO L, MgO L, MgO MgO MgO MgO

Ol 0·001 0·001 0·001 0·001 0·0044 0·443 0·0424 0·119 0·015 0·016 0·018 0·022 0·022 0·0944

Mt

T2 T2 T2 T2

0·025 0·167 0·029 0·024 0·0637 0·063 0·530 1·054 0·0005 0·001 0·002 0·0006 0·0003 0·0016

Ilm

T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2

0·00372 0·0255 0·0255 0·0724 344 67 0·0393 0·0603 4·47 6·39 8·30 0·0580 1·304 41 wt%

Ap SiO Av. set=DK SiO2 MgO MgO SiO2 SiO2 L, SiO2 L, SiO2 L, SiO2 Av. SiO2

0·005 0·005 0·006 0·005 62 800 189 44·4 26·6 23·5 20·0 0·06 0·03 2·4

Zirc

Table 1 Example of one of the mineral/liquid partition coeﬃcient datasets used in this paper, corresponding to an An50 plagioclase. Cpx=clinopyroxene; Plag=plagioclase; Opx=orthopyroxene; Ol=olivine; Cte=chromite; Mt=magnetite; Ilm=ilmenite; Ap=apatite; Zirc=Zircon. Columns to the right of the D values record the way that D was calculated, or gives the source of the data or equations used. For example, ‘29, An’ implies that D values were calculated from plagioclase An-content using equation #29. Plag Ds were calculated using equations in Be´dard 2006a. Opx and Ol Ds were calculated using equations in Be´dard 2007 and Be´dard 2005, respectively. Values for Ds in Cpx and Ap are from mss in preparation. Values for chromite, magnetite and ilmenite are from the compilation of Be´dard 2006b unless otherwise speciﬁed, or were calculated from the constants given in Be´dard et al. 2009. ‘MgO’ implies that D values were calculated from the melt MgO content, itself calculated from the plagioclase An-content (MgOwt% of melt=An (molar fraction)14·066+0·3663; see text). For An=0·5, then MgO=7·4%. Melt FeO and SiO2 contents were assumed constant at 11% and 55%, respectively, and so in this case MgO#=MgO/(MgO+FeOt)=0·402. The tetrahedral Al-contents of Opx and Cpx (Aliv) are assumed constant at 0·05 and 0·02 respectively. For Plag, MSAn=calculated from MgO, SiO2 and An-contents using multiple regression analysis. SAn=calculated from SiO2 and An-contents using multiple regression analysis. Av.=average of experimental and natural data. L means that the Ds for the rare-earth elements and Y were calculated using the lattice strain model of Blundy and Wood (1994) from the variable associated with it (e.g., ‘135–7, L, MgO’ means that the D value was calculated using equations 135 to 137, using the Lattice strain model calibrated against the MgO content of the melt). For Ap, AvSM means that the value of D is the average of Ds calculated from regressions against MgO and SiO2. HD=calculated using the equation of Hart & Davis (1978: DNi=124/(MgO)0·9). Values of DZircon Ga, Ni, Cu, Zn and V are unknown and were set=1.

80 JEAN H. BEDARD

DNd DSm DZr DHf DTi DEu DGd DTb DDy DY DHo DEr DTm DYb DLu DGa DCr DCo DNi DCu DZn DV DSc

0·283 0·442 0·0214 0·0466 0·178 0·0302 0·563 0·606 0·631 0·638 0·638 0·630 0·612 0·587 0·560 0·339 4·32 1·61 0·796 0·164 0·621 0·388 0·934

Table 1 Continued.

L, Aliv L, Aliv MgO MgO MgO MgO L, Aliv L, Aliv L, Aliv L, Aliv L, Aliv L, Aliv L, Aliv L, Aliv L, Aliv MgO MgO MgO MgO Aliv SiO2 MgO MgO

Cpx

0·105 0·0885 0·0184 0·0537 0·110 0·275 0·0731 0·0656 0·0583 0·0540 0·0519 0·0464 0·0417 0·0378 0·0344 0·964 0·116 0·165 0·440 0·00017 0·0112 0·0436 0·00148

64, L, An 64, L, An 34b, An 40a, SAn 8, An 46b, An 64, L, An 64, L, An 64, L, An 64, L, An 64, L, An 64, L, An 64, L, An 64, L, An 64, L, An 25a, MgO 23, An 29, An 24, An 30, MSAn 31, MgO 27, An 28, MgO

Plag 0·0104 0·0223 0·0156 0·0354 0·172 0·0184 0·0400 0·0521 0·0666 0·0769 0·0824 0·0988 0·115 0·132 0·147 0·186 8·25 1·87 1·95 0·0671 1·22 0·385 0·858

135–7, L, MgO 135–7, L, MgO 31, MgO 32, MgO 15, MgO 42a, MgO 135–7, L, MgO 135–7, L, MgO 135–7, L, MgO 135–7, L, MgO 135–7, L, MgO 135–7, L, MgO 135–7, L, MgO 135–7, L, MgO 135–7, L, MgO 95, MgO 79, MgO 84, MgO 87, MgO 94, Aliv 88, MgO 91, MgO 81, MgO

Opx 0·00070 0·00240 0·0180 0·0110 0·0330 0·00841 0·00614 0·00939 0·0140 0·0176 0·0197 0·0264 0·0340 0·0420 0·0502 0·103 1·037 4·21 15·9 0·11 1·42 0·15 0·238

L, MgO L, MgO MgO MgO MgO MgO L, MgO L, MgO L, MgO L, MgO L, MgO L, MgO L, MgO L, MgO L, MgO Av. MgO MgO HD Av. MgO Median MgO

Ol 0·026 0·024 0·0631 0·042 0·837 0·025 0·018 0·019 0·018 0·018 0·018 0·018 0·018 0·018 0·018 3·20 63·7 5·85 19·65 0·114 4·24 0·104 0·656

Mt

T2 T2 T2 T2 T2 T2 T2 T2

T2 T2 T2

0·0036 0·0093 0·218 0·356 51 wt% 0·0004 0·0188 0·0258 0·0346 0·0409 0·0443 0·0546 0·0652 0·0755 0·0853 0·0863 8·89 1·05 9·72 3·16 0·214 10·34 0·859

Ilm

T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2 T2

T2 T2 T2 T2

9·83 10·64 123 506 0·0465 5·59 9·18 7·94 6·55 5·69 5·27 4·18 3·30 2·61 2·08 0·280 9·0 0·17 0·0468 0·28 1·79 0·022 0·05

Ap L, SiO2 L, SiO2 MgO MgO Av. AvSM L, SiO2 L, SiO2 L, SiO2 L, SiO2 L, SiO2 L, SiO2 L, SiO2 L, SiO2 L, SiO2 set=DCu Av. Av. set=DFeO Av. Av. set=DAl AvSM

21·7 17·7 66 wt% 2640 0·056 12·1 15·0 37·3 60·0 80·0 120 200 300 490 632 1 9 16 1 1 1 1 57·5

Zirc GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS 81

82

JEAN H. BEDARD

from the same intrusion. Where a cumulus origin is inferred for magnetite, ilmenite or apatite, more complex oxide–apatite melting modes were used to calculate the residual mode. A cumulus proportion of 0·1 was assumed for both magnetite and ilmenite, and 0·08 for apatite, with the proportions of the better-known silicate phases being reduced by dilution. Other assumptions and detailed procedures involved in the modelling were discussed at length in Be´dard (1994, 2001) and are not repeated here.

2.2. Data sources Modern whole-rock trace element data were gleaned from the literature (Labrieville in Owens & Dymek 2001; Mattawa in Owens & Dymek 2005; Raudot, DeLaBlache and Rivie`re Pentecoˆte in Francis et al. 2000), and from Que´bec Government studies supplied through the courtesy of Daniel Lamothe and Claude He´bert (Ministe`re des Resources Naturelles et de la Faune du Que´bec), who extracted all available anorthositerelated whole-rock data from the SIGE u OM database. Older data that showed spiky normalised trace element patterns were discarded. An anorthosite from the Mealy Mountains complex of Labrador (sample L5) was collected during a ﬁeldtrip led by C. Gower in 1989, and was analysed (Table 2) as described in Be´dard (2001).

3. Results of inverse models 3.1. Conventions, and the presence of positive Zr–Ti–P peaks

Figure 2 Error analysis applied to a Vanel anorthosite (441744, with a few elements interpolated from other rocks), showing consequences of varying. (a) An-content from An50 to An70 (the rock analysis is also shown); (b) the assumed trapped melt fraction from 0·04 to 0·1; and (c) plagioclase/melt D values by 1 sigma, from the equations in Be´dard 2006a. Values in this and forthcoming graphs are normalised to N-MORB of Sun & McDonough 1989, except K (600 ppm), U (0·047 ppm) and Th (0·12 ppm) as per Jochum et al. 1983, and compatible elements which are modiﬁed from Pearce & Parkinson 1993 by comparison with compiled data: Ga (16 ppm), Cr (275 ppm), Co (47 ppm), Ni (135 ppm), Cu (100 ppm), Zn (83 ppm), V (250 ppm), Sc (40 ppm). Elements where no error bars are shown were not computed from the An-content.

cumulate rocks (see table e2 in Be´dard 2001). The point of disappearance of minor silicate, phosphate or oxide phases during backstripping was used to constrain the TMF in most models calculated in the present paper. For most anorthosites, an assumed TMF of about 5–10% completely eliminates all phases except feldspar. Most maﬁc cumulate models were reduced to simple two- or three-phase assemblages between 10 and 15% TMF. In a few cases, the TMF was increased beyond this point to bring models into congruence with other models

The modelled rocks were subdivided both by intrusion and lithology. Lithological subtypes include anorthosites (ss) with >90% plagioclase (normative), leuco-gabbroids with 90– 65% plagioclase, gabbroids with 65–40% plagioclase, melagabbroids with 40–30% plagioclase, pyroxenites with <30% plagioclase and >50% pyroxene, and peridotites with >50% olivine. Average values for each suite are shown in Table 3. The input data, calculated norms, and individual models are available as an electronic Appendix (Table A1 – Supplementary Material). Many inverse models yielded normalised melt proﬁles with positive Ti–P–Zr anomalies (vs. adjacent REE). There are 18 models (including some calculated from plagioclase megacrysts) which have positive Zr anomalies, which are too large to be due to use of an overly low D value, or to analytical imprecision. They may reﬂect a nugget eﬀect caused by analysis of samples which are too small to include a representative proportion of post-cumulus zircon. Alternatively, some of the analysed rocks and feldspar megacrysts may contain xenocrystic zircon. Inherited zircon occurs in other anorthosite provinces (e.g. Scha¨rer et al. 1996) and may also be present in Grenvillian anorthosites. Since these zircons contribute to the Zr-budget, addition of zircon to the residual mode can erase this signature. Most samples require 0·001% to 0·002% zircon, with a maximum of only 0·013%. Positive P-anomalies are seen in 22 model melts (e.g. Fig. 3). In many cases, very small changes in the abundance of residual apatite smooth out the proﬁle. The simplest explanation is therefore that the modal proportion of apatite used during backstripping (8%) diﬀered from the real proportion of crystallising apatite in some cases. Similarly, positive Ti anomalies occur in 35 models (e.g. Fig. 3). Some may reﬂect small diﬀerences between model and real proportions of ilmenite in the crystallising melt. However, in some examples, extremely large proportions of residual ilmenite are needed to drag down the Ti peak (up to 13%). A primary positive Ti and/or P

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

83

Table 2 Analysis of L-5 anorthosite from the Mealy Mountains massif of Labrador. Major elements and Sr, Cr, Ni, Cu, Zn, and V by ICP-ES, with all other trace elements by ICP-MS. Data generated at the Institut National de la Recherche Scientiﬁque – Eau, Terre et Environnement, in Que´bec City, with methods and accuracies as discussed in Be´dard 2001. Major elements (wt%) SiO2 TiO2 Al2O3 Fe2O3 MnO MgO CaO Na2O K 2O P2O5 LOI Total CIPW Norm

54·14 0·14 28·11 1·06 0·02 0·61 10·78 4·47 0·83 0·02 0·79 100·97

Anhydrous Fe2+ /Fet=0·9 Quartz Corundum Orthoclase Albite Anorthite Hypersthene Magnetite Chromite Ilmenite Apatite An/An+Ab+Or (molar) 54·2 Residual Mode Plag 90% TMF 10%

0·42 0·31 4·90 37·79 53·31 2·81 0·15 0·01 0·27 0·05

Trace elements in rock (ppm)

Cs K Rb Ba Th U Nb Ta La Ce Pr Sr P Nd Sm Zr Hf Ti Eu Gd Tb Dy Y Ho Er Tm Yb Lu Cr Ni Cu Zn V Sc

anomaly in the melt is unlikely, and these peaks are most plausibly explained by either: heteradcumulus growth of these phases, or post-cumulus metasomatism by a Ti–P-enriched immiscible melt (e.g. Philpotts 1966, 1976; Veksler et al. 2007). Adding small amounts of apatite or ilmenite to the residual mode can correct for these eﬀects. The adjustment for ilmenite has only a very small eﬀect on the calculated abundances of most trace elements, but adjustment of modal apatite and zircon can have signiﬁcant impact on calculated REE (Fig. 3). Nevertheless, if no corrections were made, the basic conclusions of the present paper would remain identical.

3.2. Labrieville massif The Labrieville alkalic anorthositic pluton (ca. 300 km2) was emplaced at 1·01 Ga (Owens et al. 1984). The Labrieville models are considered at some length to illustrate the methodology. This pluton has been divided by Owens & Dymek (2001) into an inner core sodic anorthosite (An<32), an outer core anorthosite that is more calcic (An>32), and a leucogabbroic marginal/border facies. Paradoxically, the sodic core has the most magnesian pyroxenes, whilst the calcic margin

0·1352 6890 6·24 603 0·0094 0·0272 0·386 0·01 2·47 4·76 0·656 1325 87·3 2·39 0·488 1·58 0·042 839 0·608 0·899 0·037 0·198 0·801 0·035 0·084 0·010 0·061 0·009 35 30 26 18 3 3

Model melt (ppm)

Cs K Rb Ba Th U Nb Ta La Ce Pr Sr P Nd Sm Zr Hf Ti Eu Gd Tb Dy Y Ho Er Tm Yb Lu Cr Ni Cu Zn V Sc

0·7925 22985 43·5 1207 0·052 0·177 2·32 0·073 12·5 24·5 3·49 528 464 13·2 2·92 14·4 0·305 4354 1·73 5·81 0·248 1·38 5·75 0·251 0·629 0·080 0·477 0·0740 191 82·9 260 172 19·7 25·7

has the least magnesian pyroxenes. The pluton also contains a hemo–ilmenite/nelsonite deposit (Dymek & Owens 2001; He´bert et al. 2005). Most Labrieville data are from Owens & Dymek (2001). Although these analyses lack U and many REE, they are useful because they include data on plagioclase megacrysts which are (a priori) single crystals. As such, their TMF must be=0, unless disequilibrium crystallisation allowed incorporation of abundant melt inclusions (e.g. Be´dard 2001). Assuming that these megacrysts were originally at equilibrium, then models derived from the megacrysts assuming TMF=0 can be compared to models calculated from whole-rock data on cumulates (composites of minerals + trapped melt), to constrain the TMF for this suite of rocks. Figure 4a shows the NMORB-normalised (normal midocean ridge basalt) trace element proﬁles of melt calculated from inner core megacryst 285p, as compared to the model melt calculated from Labrieville inner core anorthosite 285. Both model melts are fractionated, with relative L/H REE enrichment (light and heavy rare earth elements, respectively), LILE-enrichment (large ion lithophile elements), prominent positive Sr–Eu anomalies, a negative Th-anomaly, a spiky

Cs K Rb Ba Th U Nb Ta La Ce Pr Pb Sr P Nd Sm Zr Hf Ti Eu Gd Tb Dy Y Ho Er Tm Yb Lu Ga Cr Co Ni Cu Zn V Sc

1·438 13370 31·3 289 1·49 1·04 7·4 0·993 9·46 17·3 2·09 2·90 252 517 8·56 1·58 34·4 0·930 3826 1·842 1·512 0·196 0·975 5·96 0·196 0·581 0·079 0·559 0·113 18·1 178 40·7 55·1 32·7 37·8 130 14·7

Av.

1·127 5814 27·5 105 1·36 0·53 6·4 0·300 3·01 5·22 0·56 6·65 54·8 234 2·32 0·45 23·4 0·427 1207 0·561 0·552 0·074 0·343 2·64 0·106 0·255 0·025 0·295 0·051 2·5 282 23·6 50·3 30·3 34·2 98 8·4

Std

Low-Sr P–SAS (31)

16 31 28 31 9 9 10 5 30 30 30 22 31 26 30 29 25 3 31 30 30 29 30 29 11 24 7 23 16 28 28 19 22 16 17 16 18

N

0·186 45700 39·3 1095 0·096 — 6·8 0·777 13·6 25·3 2·94 3·11 480 1046 12·5 2·08 58·1 1·460 3702 2·94 1·506 0·181 1·164 3·66 0·247 0·438 0·086 0·343 0·072 22·8 19·4 9·59 4·1 21·8 59·6 149 6·84

Av.

0·127 9327 11·9 220 0·015 — 5·0 0·814 3·38 5·88 0·85 3·54 53·5 531 4·47 0·557 19·4 1·011 1320 0·672 0·743 0·074 0·435 1·76 — 0·217 — 0·112 0·041 1·6 28·9 9·39 2·4 16·1 33·7 64 4·83

Std

High-Sr P–SAS (13)

5 13 13 13 4 0 3 10 13 13 4 5 13 13 4 13 12 10 13 13 4 13 4 4 1 3 1 9 6 13 11 12 4 3 8 5 12

N

2·27 12690 64·8 268 2·87 1·46 14·1 1·71 12·4 23·2 3·01 1·59 279 698 12·9 2·68 75·0 2·525 6346 2·193 2·90 0·487 2·44 13·8 0·496 1·45 0·175 1·29 0·212 22·3 118 54·1 56·2 88·6 109 223 25·3

Av. 2·61 6895 74·5 133 3·020 1·079 18·3 1·89 3·42 6·51 0·88 0·38 57·4 301 4·07 1·00 43·8 1·285 2788 0·879 1·257 0·208 1·13 5·9 0·242 0·66 0·129 0·61 0·108 3·6 68 22·8 42·7 73·5 90 178 9·7

Std

Low-Sr FAGS (20)

8 14 11 14 7 4 11 8 14 14 14 6 12 11 14 14 14 8 14 14 14 14 14 11 13 14 10 14 13 10 10 10 14 6 9 7 6

N 0·236 36062 33·0 921 0·183 — — 1·24 17·3 35·7 — 7·12 498 1784 — 4·36 126 2·707 8514 3·30 — 0·557 — 40·5 — — — 1·46 0·227 23·0 12·2 30·2 20·8 127 188 162 24·9

Av. — 2430 7·0 116 0·042 — — 1·19 4·99 11·2 — 0·20 38·3 730 — 1·94 73·8 1·824 2235 0·648 — 0·321 — — — — — 1·23 0·199 2·2 5·9 11·6 7·1 47 67 11·7 15·6

Std

High-Sr FAGS (5)

1 5 5 5 5 0 0 5 5 5 0 2 5 5 0 5 5 5 5 5 0 5 0 1 0 0 0 5 5 5 5 5 4 2 5 3 5

N — 7414 — 350 — — — 1·55 6·47 10·7 1·28 — 338 286 4·98 0·960 24·0 — 2211 1·22 1·06 0·217 1·25 12·6 0·384 1·11 0·206 1·50 0·272 — 97·8 — 80·3 — — 262 —

Av. — 1724 — — — — — — 2·75 4·05 0·52 — — 53·3 1·85 0·304 — — 399 0·59 0·32 0·034 0·26 — 0·036 0·10 0·044 0·029 0·025 — 9·4 — 11·7 — — — —

Std

U (2)

0 2 0 1 0 0 0 1 2 2 2 0 1 2 2 2 1 0 2 2 2 2 2 1 2 2 2 2 2 0 2 0 2 0 0 1 0

N 3·55 21480 83·8 373 5·14 3·30 9·0 1·99 33·5 65·9 7·70 1·91 247 1733 30·7 6·53 161 5·13 7248 2·75 4·88 0·766 4·42 29·2 0·964 2·45 0·664 2·81 0·409 20·7 130 22·9 30·0 42·9 67 132 18·9

Av. 4·36 9693 76·6 153 7·31 2·22 6·1 1·76 18·3 33·3 4·02 1·00 29 942 16·1 2·80 168 2·36 5160 1·19 2·99 0·564 3·14 15·8 0·576 1·49 0·342 1·54 0·247 2·2 160 19·6 33·5 32·3 40 113 8·9

Std

Low-Sr E–SAS (12)

10 12 12 12 9 7 9 8 12 12 11 11 12 10 12 11 12 7 12 12 11 12 11 11 10 11 3 11 12 12 10 11 12 11 12 11 11

N — 26300 30·4 940 3·57 1·614 8·0 1·29 40·9 80·0 9·33 0·858 501 3269 46·1 8·35 245 5·90 12055 3·75 7·18 0·898 4·57 24·7 0·938 2·39 0·426 2·23 0·316 24·1 158 17·9 16·0 26·9 88 139 17·6

Av. — 7857 17·1 448 3·65 1·183 3·6 0·81 15·2 30·8 2·87 0·598 126 1411 14·1 2·45 184 1·31 6807 1·21 2·37 0·268 1·30 7·11 0·302 0·81 0·083 0·65 0·168 3·6 312 7·3 9·7 11·8 43 64 7·6

Std

High-Sr E–SAS (7)

0 9 8 9 4 2 6 3 9 9 7 7 9 8 8 9 9 6 9 9 7 7 7 8 6 7 4 9 6 9 7 8 7 8 8 8 8

N 3·54 20017 30·7 695 4·17 0·518 15·8 2·34 29·4 57·5 8·24 2·45 201 2860 37·0 9·61 99 1·71 16820 3·98 10·46 1·394 8·88 47·0 2·16 5·13 0·874 4·48 0·724 23·5 162 42·7 76 267 151 355 40·5

Av. 2·27 8783 17·2 395 12·2 0·174 7·2 2·20 8·95 16·4 1·78 — 41 997 11·6 2·39 107 2·21 7854 1·35 5·11 0·623 4·89 20·9 1·02 2·88 0·414 1·80 0·309 4·4 220 23·3 84 408 86 233 16·7

Std

Low-Sr ES (16)

8 16 14 16 12 3 15 7 16 16 5 1 15 14 15 16 15 13 16 16 6 14 5 16 9 5 7 16 14 13 8 12 9 6 6 5 13

N 1·275 33290 29·7 1105 0·458 — 15·8 1·97 51·3 108 20·4 0·788 518 5725 101 15·2 274 4·77 18270 5·61 18·0 1·963 13·5 49·7 2·56 5·71 0·816 3·90 0·627 28·5 37·7 29·9 7·3 52·4 174 224 22·3

Av. — 2178 8·98 138 0·244 — 4·48 0·61 17·8 33·5 — — 64 2647 — 4·22 73 1·36 2274 0·32 — 0·468 — 15·4 — — — 0·65 0·126 2·7 35·7 11·8 3·1 8·8 71 34 11·4

Std

High-Sr ES (4)

1 4 4 4 3 0 3 3 3 3 1 1 44 4 1 3 4 3 4 3 1 3 1 4 1 1 1 3 3 4 2 3 3 2 4 4 3

N

2·226 81675 213 5595 4·76 3·53 30·6 2·84 315 611 66·7 1·258 73 9794 316 56·1 969 19·9 24566 13·8 40·8 3·53 31·5 99·6 4·28 14·7 2·39 7·33 1·093 31·2 56·5 23·3 4·7 140·6 166 468 58·8

Av.

1·089 37161 57 3307 4·58 1·97 24·6 0·54 143 276 — — 59 2540 118 15·6 777 18·0 11440 5·53 — 2·48 — 56·1 1·69 — — 3·85 0·580 4·9 46·3 0·81 — — — — 3·0

Std

VES (3)

3 3 3 3 3 2 3 2 3 3 1 1 3 3 3 3 3 3 3 3 1 3 1 3 2 1 1 3 2 3 3 2 1 1 1 1 3

N

Table 3 Average compositions of model liquids for the diﬀerent Suites. PSAS and ESAS are primitive and enriched steep anorthositic suites, respectively. FAGS is the ﬂatter anorthositic-gabbroic suite, ES and VES are the enriched and very enriched suites, respectively. The number in brackets is the number of rocks attributed to this suite. Av.=average; Std=Standard deviation; N=number of data.

84 JEAN H. BEDARD

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

85

Figure 3 Inverse models for a Labrieville ilmenite-anorthosite 126 (Owens & Dymek 2001) are shown for ca. 20% and ca. 8% trapped melt fractions (TMF). The residual mode is shown, with the ‘adj.’ or adjusted mode resulting from the addition of residual apatite, ilmenite and zircon to eliminate positive P–Ti–Zr peaks. These adjustments also impact on the concentration of HREE, Hf and Th, elements compatible in zircon and apatite. Only minor eﬀects are seen on the model concentration of other elements, however. Also shown for comparison is the average adjusted model melt from Labrieville core anorthosites (grey squares; see Fig. 4b), and the adjusted model melt from marginal gabbroic feldspar megacryst 74p (see Fig. 4c). Note that the 126 models show an overall decrease of incompatible element abundances as the assumed TMF increases. The average PSAS anorthosite (grey squares) shows a close match to the 8% TMF models, but the 20% TMF models are too low in M–HREE Zr–Hf–Ti. Conversely, the 74p models are too enriched in M and HREE–Ti–Zr to match the adjusted 8% TMF models.

Zr–Hf–Ti segment, high Ga and low Cr. The similarity implies a cosanguineous relationship. The whole-rock anorthosite model has positive Zr–Ti spikes, while the megacryst model also shows a positive P-spike. These can be corrected by adding modest amounts of ilmenite, apatite and zircon to the residual assemblage (labelled adj. (adjusted) on the Figures). This implies that the megacryst contains xenocrystic zircon, apatite and ilmenite as inclusions. In comparison with the adjusted megacryst model melt, the adjusted whole-rock model melt calculated for a TMF of 3% (anorthosite residue) has slightly lower P–REE–Zr–Hf, but there is an almost perfect match for K–Rb–Ba–Sr–Ti–Eu–Ga. The slight diﬀerence in P–REE–Zr–Hf could indicate that the melt from which the rock crystallised was slightly less evolved than the melt from which the megacryst crystallised. In a broader sense, both models (megacryst and whole-rock) for sample 285 are similar to model melts from other Labrieville core anorthosites (Fig. 4b), implying that all belong to a common suite. Model melts from other Labrieville inner core anorthosites are shown on Figure 4b. The %TMF was ﬁxed at the point where the residual assemblage became pure anorthosite in all cases. Most samples show positive Ti spikes, and half have positive Zr spikes. The non-adjusted values for sample 108 are shown for comparison. All model melts resemble models derived from megacryst 285P and anorthosite 285 (Fig. 4a), and from ilmenite–anorthosite 126 (Fig. 3); and probably belong to a cosanguineous suite, which is deﬁned here as the primitive steep anorthositic suite (PSAS). This refers to the steepness of the normalised trace element proﬁle and the modest enrichment of LREE and MREE (middle REE). On the basis of major and trace element trends, Owens & Dymek (2001) inferred that Labrievielle rocks contain very low TMF. The inverse models presented in Figure 4b, validated by comparison with the megacryst model (285p, Fig. 4a), cor-

roborate this inference. This conclusion can guide the choice of whether high-TMF or low-TMF models should be preferred in what follows. Figure 4c examines model melts calculated from feldspar megacryst 74p and its host gabbro 74, which are located in the marginal zone of the pluton. The ﬁgure also compares these models to the average inner core anorthosite PSAS model melt, and a model melt from a marginal anorthosite (442007). The model melt calculated from megacryst 74p (diamonds) shows some resemblance to average inner core PSAS model melts (grey squares), but has a shallower HREE proﬁle, and so may not be strictly cosanguineous with this suite (in accord with the proposals of Owens & Dymek 2001). Because of the shallower HREE proﬁle and lower L/H REE ratio, this suite will be referred to as the flatter anorthositic-gabbroic suite (FAGS). The contrast between PSAS and FAGS is subtle, and there may be a continuous variation between the two suites. Model melts calculated for outer core samples (36, 107 and 218) closely resemble models for megacryst 74p (Fig. 5a) and can also be classed with the FAGS. The host to megacryst 74p is an ilmenite-rich gabbro (74) from the margin of the intrusion. Several model melts from this gabbro are shown (5, 10, 37% TMF), all adjusted slightly to eliminate positive Zr–Ti–P spikes. Note that small changes in the assumed TMF (e.g. from 5% to 10%) have modest impacts on the trace element contents and slope of the model melt proﬁles. Signiﬁcant changes in trace element content of model melts are only seen for large variations in the assumed TMF (e.g. to 37%). If such a high (37%) TMF is chosen, then the model melt for gabbro 74 approaches values for models calculated from megacryst 74p for the REE–Zr, suggesting that it might have crystallised from a melt cosanguineous with the megacryst (FAGS). However, the 37%TMF model melts from gabbro 74 are too low in K–Rb, and TMF>40% are

86

JEAN H. BEDARD

abyssal intrusions, and so this high-TMF model seems implausible. A much simpler explanation is provided through comparison with a model melt calculated from an anorthosite (442007) located at the margin of the Labrieville pluton. Values of TMF of 5–10% for gabbro 74 yield enriched model melts that closely resemble 13%TMF model melts calculated from anorthosite 442007 (except for Co–Ni–Zn–Sc); suggesting that gabbro 74 crystallised from a relatively enriched melt similar to the one that generated anorthosite 442007. If this is correct, then the relatively depleted nature of the melt in equilibrium with megacryst 74p implies that this feldspar crystal must have formed from a depleted melt and was entrained into the enriched melt from which gabbro 74 crystallised, but did not fully equilibrate with it. Other samples (89, 153, 74, 99) from the marginal zone yield similar solutions, and are now deﬁned as the enriched suite (ES, Fig. 5b). The diﬀerences between the model melt trace element proﬁles of the Labrieville complex (Figs 3–5) imply that the margin (ES), outer core (FAGS), and inner core (PSAS) of the intrusion were not strictly cosanguineous. The origin of the core-to-margin trace element enrichment (PSAS to ES) is problematic, since it is coupled to outwardly increasing An-content, with the most enriched marginal samples having the highest An-contents. Owens & Dymek (2001) favoured a pressure eﬀect to explain the zoning, but also speculated that this may be due to progressive buildup of H2O during diﬀerentiation. The origin of this enrichment will be discussed below. Owens & Dymek (2001) also inferred that Labrieville rocks crystallised from unusually alkalic melts. The models developed in the present paper support this conclusion, since the average core anorthosite model melt (PSAS) contains 6·1wt% K2O, 43 ppm Rb, 1181 ppm Ba and 478 ppm Sr; the average FAGS model melt has 4·4wt% K2O, 34 ppm Rb, 919 ppm Ba and 509 ppm Sr; and the average ES model melt has 4 wt% K2O, 30 ppm Rb, 1104 ppm Ba and 518 ppm Sr. The model K2O contents are imprecise, since D K is poorly controlled in ternary feldspars, but the model Sr and Ba contents are robust.

3.3. Lac St. Jean and Vanel complexes

Figure 4 (a) Inverse models for a Labrieville anorthosite 285 and plagioclase megacryst 285p (Owens & Dymek 2001). Models are shown with and without added apatite, ilmenite and zircon. These phases were added to the residual assemblage to eliminate positive P–Ti–Zr peaks. Only models that diﬀer from the unadjusted values are shown. Note the close resemblance of megacryst (285p) and wholerock (285) models, suggesting a cosanguineous relationship, and that both resemble the average core anorthosite model melts (PSAS). (b) Inverse models for a series of Labrieville core anorthosites 285 (Owens & Dymek 2001). Only the adjusted models are shown. All are similar and, together with sample 126, were averaged to yield the average core anorthosite model melt (PSAS) shown in the other ﬁgures. (c) Inverse models for a Labrieville ilmenite–gabbro 74 and plagioclase megacryst 74p (Owens & Dymek 2001). Adjusted models are shown, with added apatite, ilmenite and zircon. Note that the megacyst models are too enriched in HREE to match the average core anorthosite model. Three models for gabbro 74 are shown, at 5%, 10% and 27% TMF. Note that the 37% TMF model approaches but does not attain the megacryst models; whilst the 5–10% TMF models closely resemble the evolved model melt calculated from the marginal anorthosite (sample 442007).

needed to obtain a close match for the REE. Rocks with such high TMF are rarely found in large plutons (Be´dard et al. 2003a), being more characteristic of diabasic-textured hyp-

The giant Lac St. Jean anorthosite complex (>20 000 km2) was emplaced between 1160 and 1140 Ma (Higgins et al. 2002; He´bert & van Breemen 2004a). The eastern border of this massif has recently been attributed to a distinct, but roughly coeval (1180–1160 Ma) massif called the Vanel complex (He´bert & van Breemen 2004b; He´bert et al. 2009). There are 16 analyses of the Lac St. Jean complex and 24 of the Vanel complexes in the SIGE u OM database. The assumed TMF of most of these was ﬁxed where the residual mode became anorthositic. Nine anorthosites from the Vanel complex yield steeply fractionated model melts (Fig. 6a) that are grouped together as the Vanel primitive steep anorthositic suite (PSAS). In contrast to the Labrieville PSAS, Vanel PSAS have lower overall alkali contents (Av. model K2O=1·42 wt%, Rb=21 ppm, Ba= 248 ppm, Sr=270 ppm), but are otherwise quite similar. Ten Vanel rocks (most anorthositic) yield model melts that have a shape very similar to the PSAS models (Fig. 6b), but with higher overall abundances of incompatible elements, and slightly lower L/H REE. This suite will be referred to as the Vanel enriched steep anorthositic suite (ESAS). Several Vanel rocks yielded PSAS or ESAS-type model melts with markedly higher contents of Sr, Ba and K. These signatures are characteristic of the nearby Mattawa intrusion (see section 3.9) and these samples may represent oﬀshoots from its younger intrusion (C. He´bert, pers. comm. 2007). Two Vanel rocks yielded model liquids with trace element proﬁles that are ﬂatter and

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

87

Figure 5 (a) Inverse models for a series of Labrieville primitive gabbroids from the outer core of the complex (Owens & Dymek 2001). Only adjusted models are shown. Note the close similarity of all models, that they resemble the model melt from megacryst 74p, and that they have shallower L/H REE slopes than the core anorthosite PSAS melts. The evolved suite gabbroid model melt (ES) from (b) is shown. (b) Inverse models for a series of Labrieville evolved gabbroids from the margin of the complex (Owens & Dymek 2001). Only adjusted models are shown. Note the close similarity of all models, and that they resemble the model melt from sample 442007. The average primitive outer core gabbroid model melt from (a) is shown (FAGS).

slightly less enriched overall than the Vanel ESAS (Fig. 6c). They most closely resemble the Labrieville FAGS (Fig. 6c), and so these two rocks are tentatively classiﬁed as a Vanel FAG suite. A single Vanel rock (441881) yielded an ES-type proﬁle (see Fig. 7b). Among the Lac St. Jean rocks, three anorthosites yielded steep fractionated trace element proﬁles very similar to Vanel PSAS model melts (Fig. 7a). This suite probably constitutes much of the Lac St. Jean complex, but is under-represented in the dataset. Twelve rocks from the margin of the Lac St. Jean complex contain abundant modal ilmenite and/or apatite, and yield model melt proﬁles characterised by overall enrichment and a ﬂat proﬁle shape (Fig. 7b). These will be referred to as the Lac St. Jean enriched suite (ES) henceforth. Most are leucogabbroids, but three have diﬀerent modes, one being a mela-gabbronorite (442409), another an ilmenite harzburgite (442421), and a third being anorthositic (441881). Nonetheless, despite wide variation in residual modes, all yield very similar

model melts. The pattern for ES model melts is commonly very spiky, with negative Zr, Hf, Ti and Sr troughs. Note that these proﬁles were not adjusted by adding ilmenite, and so the Ti and P-peaks have signiﬁcance. One Lac St. Jean rock (237171) yielded a very enriched model trace element proﬁle that does not appear to ﬁt in with the ES, and which is classed as the very enriched suite (VES, see Fig. 9c). This VES model has many of the same troughs and peaks, as do the ES models, but has a more fractionated L/H REE proﬁle and extremely high alkali contents similar to those of the Mattawa intrusion VES. It may represent an oﬀshoot from the Mattawa intrusion.

3.4. Vallant complex The Vallant complex is undated. Twelve anorthositic to troctolitic rocks from this complex yield steeply fractionated PSAS type proﬁles that closely resemble Lac St. Jean and Vanel PSAS models in most respects (Fig. 8a). Four yielded extremely high model Cu (up to 15713 ppm Cu for sample

88

JEAN H. BEDARD

Figure 6 Inverse models for a series of Vanel complex anorthosites and leucotroctolites that deﬁne the primitive (a) and enriched (b) steep anorthositic suites (PSAS and ESAS, respectively), and (c) the ﬂatter anorthositicgabbroic suite (FAGS).

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

89

Figure 7 (a) Inverse models for Lac St. Jean rocks yielding primitive steep anorthositic suite (PSAS) model melts. (b) Inverse models for Lac St. Jean rocks (and one Vanel rock) that yielded ﬂat, enriched trace element proﬁles, classed as the enriched series (ES). The ﬁeld of Vanel ESAS models from Figure 6b is shown for comparison.

625280), and Co–Ni (up to 1369 ppm Ni, and 277 ppm Co for 625280); far above the range observed in the Lac St. Jean and Vanel models. The anomalous Vallant rocks contain high S (1·44% for 625280) and the anomalous models probably contain sulphides (Corriveau et al. 2007). The Co–Cu–Zn– V–Ni averages for this intrusion do not include these anomalous rocks. Four Vallant anorthosites and troctolites yielded more enriched model melts that closely resemble Vanel ESAS models (Fig. 8b); whilst ﬁve other anorthosites and troctolites yielded ﬂatter, generally more depleted models similar to the Vanel FAGS (Fig. 8c).

margins and dykes that Francis et al. (2000) interpreted as quenched melts. Two of the troctolites provided by Francis et al. (2000), together with two leucotroctolite analyses from the SIGE u OM database, yield model melts that are similar to the average Vanel and Lac St. Jean PSAS model melt (Fig. 9a) and are classed as the De La Blache PSA Suite. Another ﬁve analyses yield slightly ﬂatter proﬁle shapes with higher overall HREE contents (not shown) that are very similar to the Labrieville FAGS average and are grouped together as the De La Blache FAG Suite (Fig. 9b). Two feldspar-poor rocks from the Blache complex yield ﬂatter enriched proﬁles (ES) (Fig. 9c).

3.5. De La Blache complex

3.6. Rivie`re Pentecoˆte intrusion

The De La Blache troctolite–anorthosite complex has been dated at 1·327 Ga by Gobeil et al. (2002), and has ﬁne-grained

The Rivie`re Pentecoˆte troctolite–anorthosite complex has been dated at 1·36 Ga (Nantel & Martignole 1991; Martignole et al.

90

JEAN H. BEDARD

Figure 8 (a) Inverse models for a series of Vallant complex rocks with steep fractionated proﬁles very similar to the Vanel/Lac St. Jean PSAS models (grey ﬁeld). Some of the samples are enriched in sulphide and yielded strong Cu–Ni–Co–Zn anomalies. (b) Inverse models for four Vallant complex rocks with more enriched and fractionated melt proﬁles very similar to the Vanel ESAS models (grey ﬁeld). (c) Inverse models for ﬁve Vallant complex rocks with ﬂatter proﬁles very similar to the Vanel FAGS models (grey ﬁeld).

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

1993). Two analyses are provided by Francis et al. (2000). Troctolite PC31 (An63) yields a 5%TMF model melt that closely resembles the average Lac St. Jean and Vanel PSAS models (Fig. 9a), and is classed as a PSAS-type rock.

3.7. Havre St. Pierre complex The Havre St. Pierre anorthosite is dated at ca. 1·062 Ga (van Breemen & Higgins 1993). The Rivie`re Romaine (RR in Fig. 1) and Lac Fournier (LF in Fig. 1) lobes may be related, but are undated. The Atikonek massif (1·13 Ga) appears to be slightly older, but the date is from an associated granite (Emslie & Hunt 1990). These massifs are poorly represented in the database, with only three anorthosite analyses from the SIGE u OM compilation. All three yield model melts that are similar to average Vanel ESAS, albeit with a very slightly ﬂatter shape (Fig. 9b).

3.8. Mealy Mountains anorthosite A single anorthosite analysis from the ca. 1·65–1·62 Ga (James et al. 2000) Mealy Mountains anorthosite is presented here (Table 2). The data was generated as outlined in Be´dard (2001). The inverse model is very similar to the average Lac St. Jean and Vanel PSAS model (Fig. 9a).

3.9. Mattawa complex The 1·012 Ga (He´bert et al. 2005) Mattawa andesine– anorthosite intrusion is of nearly the same age as the Labrieville intrusion, and the two may be related (Owens & Dymek 2005). It is also characterised by a predominance of andesine anorthosite and contains hemo-ilmenite deposits and inclusions of labradorite anorthosite. Six anorthosites (two ilmenite-rich) and one leuco-troctolite yielded PSAS-type model melts (Fig. 9a). Most have low An-contents and are high in Sr and LILE, and represent a high-Sr alkaline PSAS end-member similar to the Labrieville high-Sr alkaline PSAS. Others have higher An and lower Sr-contents and may represent enclaves of the adjoining Vanel anorthosite. Three anorthosites yielded ESAS type melts somewhat similar to Vanel ESAS (Fig. 9b), though with higher overall Sr-contents. Data are incomplete (Owens & Dymek 2005), but model melts from these andesine anorthosites are generally similar to the ESAS models, and they are grouped together henceforth. Two rocks yielded VES-type melts (Fig. 9c).

3.10. Raudot intrusion The undated, 300 m-thick, Raudot intrusion from the Manicouagan region contains dunitic to leucotroctolitic rocks (Francis et al. 2000), and has ﬁne-grained margins that Francis et al. (2000) interpreted to be quenched melts. The availability of near-liquid compositions has allowed the relationship between melt MgO content and feldspar An-content to be constrained (see above). Six analyses of cumulate rocks from the Raudot intrusion are available (Francis et al. 2000), and yield model melts of the FAG and ES type (Fig. 9b and c), with a dunite yielding a depleted, U-shaped model melt similar to a boninite (e.g. Be´dard 1999, Fig. 10a).

4. Comparisons and inferences about processes The average values of the diﬀerent model melt suites from each pluton or massif are compared in Figures 9 to 11. The diﬀerent lineages share many traits in common. All are LREE–LILEenriched, and most have negative Th–Nb anomalies and positive Eu anomalies. The absolute values of the Eu anomalies may not be signiﬁcant, since the D values for Eu in

91

plagioclase in the present models are under-constrained. Conversely, the prominent positive Sr-anomalies that characterise PSAS and FAGS melts are much less prominent in the other magma types (Figs 9 and 10). The Grenville PSAS averages deﬁne a homogeneous population (Fig. 9a) that is very similar to the average normal anorthosite model melt from the Nain Plutonic Suite (Fig. 11a) calculated by Be´dard (2001). The similarity suggests that this magma type is common to both the NPS and Grenville Province; and that use of constant Ds for the NPS calculations has not led to a major error. On the other hand, the small apparent enrichment in LREE and LILE in the Nain models probably reﬂects the eﬀect of the older set of D values that were used. Similarly, the FAGS averages (Fig. 9b) also deﬁne a fairly homogeneous population essentially indistinguishable from the ﬂat anorthosite model melt from the NPS (Fig. 11a). The similarity in the shape of the PSAS and FAGS proﬁles (Fig. 11a) suggests a common source. The increasing divergence for the HREE segments suggests that the petrogenesis of the steeper PSAS melts may involve larger proportions of a more HREE-compatible phase such as garnet, amphibole or clinopyroxene; or admixture with a component having higher HREE-contents. The Grenville ESAS averages seem to represent a distinct population (Fig. 10b) that has a lower L/H REE ratio and higher overall trace element abundances in comparison to the FAGS (Fig. 10a). The Grenville ES also form a fairly coherent population (Fig. 9b), that resembles the average enriched maﬁc model melt from the NPS (Be´dard 2001; Fig. 11b). The main diﬀerences are that the Grenvillean model melts have higher overall LILE–Th–Eu concentrations in comparison to the NPS maﬁc model melt. Be´dard (2001) interpreted the enriched maﬁc model melts from the NPS as possibly belonging to a continental tholeiitic lineage. Interestingly, the Grenville ESAS and ES are almost indistinguishable for elements to the left of Zr (Fig. 11b). This might imply a common source, but involvement of diﬀerent proportions of a more HREEcompatible phase such as garnet, amphibole or clinopyroxene. The Labrieville mica lamprophyres analysed by Owens & Tomascak (2002; Fig. 11b) are distinct in having slightly steeper proﬁles, but otherwise show some resemblances, suggesting some link. The VES (Figs 9c & 10a) have very steep proﬁles, not unlike those of the PSAS. Is it possible that they may represent fractionated melts derived therefrom? The two U-shaped models (Fig. 10a) may perhaps reﬂect involvement of amphibole, or genesis from refractory arc mantle. In view of the limited number of analyses available, these two subgroups (VES and U-type) are not discussed further. Figure 12a shows the calculated liquid Sr-content vs. the normative An-content of the individual rocks modelled; whilst Figure 12b shows the NMORB-normalised Sr-anomaly (in log units) calculated by interpolating to the nearest available REE. It is apparent that there are two main data clusters. The andesine-anorthosites of the Labrieville and Mattawa intrusions constitute a high-Sr/low-An group, in accord with the suggestion of Owens & Dymek (2001), who pointed out that Labrieville was unusually alkaline. In contrast, the Vallant, Blache, Lac St. Jean, Vanel, Raudot and Pentecoˆte intrusions constitute a low-Sr/high-An group. The high- and low-Sr averages are compared in Figure 10b and c. On average, the high-Sr subgroups have slightly steeper proﬁles (higher L/H REE) and are slightly enriched in Ba and P in comparison with the equivalent low-Sr suites. Models calculated from the very enriched suite (VES) rocks from Mattawa (and Lac St. Jean) plot with the high-Sr group (Fig. 12a). The single analysis from the Mealy Mountains massif is anomalous in showing high-Sr contents at high-An contents. This may represent a diﬀerence

92

JEAN H. BEDARD

Figure 9 Compares results for the diﬀerent intrusions and Suites. The number between brackets is the number of models in each average. (a) PSAS, (b) FAGS and ESAS and (c) ES and VES models.

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

93

Figure 10 (a) Compares averages for PSAS, FAGS, ESAS, ES, VES and U-Shaped models from all Grenville intrusions. (b) Compares average Low-Sr and High-Sr PSAS and ES models. (c) Compares average low-Sr and high-Sr ESAS models.

in source composition. Two data from Havre St. Pierre fall between the low- and high-Sr groups, while a third model from this intrusion plots with the low-Sr/high-An group. Finally, the two labradorite–anorthosite enclaves from Mattawa are distinct, since one is high-Sr, while the other is low-Sr. Figure 12a also shows how model melt Sr varies with changes in assumed TMF (the ticked curve 10–20–30%). The diﬀerences between the two groups is far beyond what could be expected from errors in assumed TMF or plagioclase/liquid D variations, and so this distinction into low-Sr/high-An vs. high-Sr/low-An appears to be a robust result. At Labrieville, the gradation from a PSAS-dominated core through a FAGS-dominated margin to an ES rim (Owens & Dymek 2001) is associated with an increase in An-content

(Fig. 12a). This could perhaps be misinterpreted as a rim-tocore fractional crystallisation series, were it not for the fact that: (1) the Mg# of maﬁc phases decreases from the intrusion core to its rim (Owens & Dymek 2001), as do the compatible elements in the model melts (Figs 4 and 5), the opposite of what is expected from fractional crystallisation; (2) the size of the positive Sr (Fig. 12b) and Eu anomalies increase from intrusion rim to core, precluding signiﬁcant plagioclase fractionation; and (3) ES and FAGS melts are more enriched in incompatible elements than the PSAS melts, and so are unlikely to be parental to PSAS (Figs 4 and 5). The high-Sr PSAS to ESAS melts from the Mattawa complex show a similar evolutionary pattern, and presumably share a common genetic scenario.

94

JEAN H. BEDARD

Figure 11 (a) Compares averages for Grenville PSAS and FAGS to equivalent Nain Plutonic Suite (NPS) anorthositic model melts (Be´dard 2001). (b) Compares average Grenville ES and ESAS models to average NPS maﬁc enriched model melts, and to the average Labrieville biotite lamprophyre from Owens & Tomascak (2002).

In contrast, the low-Sr model melts appear to show a simpler trend of systematically decreasing Sr, and a change from positive to negative Sr anomalies as An decreases (Fig. 12b); which would be consistent with a plagioclasedominated fractional crystallisation scenario. A series of simple equilibrium melting and fractional crystallisation models are developed in the next section to test possible evolutionary processes. A full petrogenetic model has yet to be developed, but it is hoped that these tentative ﬁrst steps can at least eliminate impossible hypotheses and rough out a plausible scenario for more rigorous evaluation in the future.

5. Fractional crystallisation models A series of simple fractional crystallisation models using constant D values (Table 1, Fig. 13) were set up to see whether it is possible for a low-Sr PSAS parent to evolve through a low-Sr FAGS or a low-Sr ESAS intermediate melt to a low-Sr ES residual melt. Figure 13a shows model residues from 60% fractionation of the average low-Sr PSAS (black circles) model melt. Residues from anorthositic (open circles) and troctolitic

(black squares) fractionation assemblages are shown. Note that the troctolitic residue models are only shown where they diﬀer from the anorthositic models. Through trial and error, reasonable matches with the target FAGS (grey squares) for the HREE were produced, but the LREE and LILE are always over-enriched in the models vs the target. Notably, the residual melt generated by anorthosite fractionation shows extreme Sr-depletion, rendering this an implausible hypothesis. The troctolitic fractionation model can improve the ﬁt for Sr, but shows excessive Ni–Co depletion, and the LREE–LILE misﬁt remains. The misﬁt in the overall pattern implies that it is implausible to derive FAGS from a PSAS parental melt by simple fractional crystallisation. Coupled assimilation– fractionation (AFC) models involving interaction with continental crust cannot resolve this problem, since the LREE–LILE ﬁt would worsen. A similar set of models test whether low-Sr ESAS (grey squares) and ES (grey diamonds) model melts can be derived from low-Sr PSAS parents (black circles, Fig. 13b). Anorthositic (open circles), and feldspathic websterite (black squares) fractionation assemblages were found to yield fairly

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

95

Figure 12 (a) Sr (ppm) of model melts, and (b) Log of N–MORB-normalised Sr anomaly in model melts (interpolated to nearest REE) vs normative An-content of rock. Note the existence of two trends.

close matches with the ESAS model melt REE at 80% crystallisation, but both fractionation assemblages lead to overenrichment of the LILE. Note that only feldspathic websterite models that diﬀer from the anorthosite models are shown. Furthermore, the residue from anorthosite fractionation shows excessive Sr-depletion, and over-enrichment in Cr–Co–Cu– Zn–Sc; whilst the residue from pyroxenite fractionation shows excessive Ni–Co depletion. Extraction of gabbroic assemblages (black diamonds) shares many of these defects and worsens the HREE ﬁt. About 20% biotite fractionation would be needed to suppress LILE-enrichment, which seems implausible. Thus, it does not seem possible to derive an ESAS residue from a PSAS parent by simple fractional crystallisation. Finally, the previously outlined fractionation models (Fig. 13b) yield reasonable ﬁts to the LREE–MREE–Sr–Ti abundances of ES model melts for 80% crystallisation; but cannot generate the requisite HREE enrichment, or generate

residual melts that are over-enriched in the LILE, and cannot account for the fact that ES model melts generally have compatible element abundances equal to or higher than is observed in the PSAS model melts. Consequently, ES melts cannot be the residua of a PSAS to ESAS to ES fractionation series.

6. Partial melting models The prominent positive Sr–Eu peaks of the model melts calculated from the Grenville anorthosites might suggest a feldspar-enriched, possibly cumulate source (e.g. Duchesne et al. 1999). This type of source might also impose an unusually aluminous nature to the melt and explain the abundance of plagioclase that eventually crystallises. Three equilibrium melting models (Fig. 14a) were calculated using depleted

96

JEAN H. BEDARD

Figure 13 Fractional crystallisation models using the constant D values of Table 1 attempting to relate the average low-Sr PSAS parental melt to (a) av. FAGS and (b) av. Low-Sr ESAS and ES model melts.

aluminous cumulate (1722A4), and an enriched aluminous and ferruginous (1722A2) gabbro–norite cumulate from the Talkeetna Arc (Greene et al. 2006) as sources. D-values used were those of Table 1, with garnet Ds from Be´dard (2006a). The open symbols represent garnet-rich residues (needed to fractionate the REE), whilst the grey diamonds shows a variant with a pyroxenitic residue. Comparison with the compositions of the average low-Sr PSAS model melt from the Grenville (grey line on Figure 14a) shows that arc cumulates are too depleted and cannot generate PSAS-type melts at reasonable degrees of melting (ca. 5–30%). The problem is compounded by the steep slope of the PSAS models. To generate a melting model with such a slope requires abundant residual garnet, which creates strongly REE-depleted melts. If smaller proportions of garnet are used (grey diamond), the models yield melts with HREE-contents that begin to approach the PSAS ﬁeld, but which have very shallow L/H REE slopes. Thus, it seems as though simple batch melting of feldspathic arc cumulates alone cannot produce PSAS or FAGS type melts. Given the trace element deﬁcit of the arc cumulate sources (Fig. 14a), a typical arc tholeiite protolith was tried. The average composition of the Mount Misery Formation, an Ordovician-age depleted arc-tholeiite sequence from Newfoundland (Be´dard 1999) was used. Figure 14a shows a

30% melting model based on this source. Although somewhat ad-hoc in terms of its residual mineral assemblage, the model yields a fairly good ﬁt to low-Sr PSAS melts. Small variations in assumed residual mode or degree of melting produce only small changes in melt composition, and so the ﬁt to PSAS is fairly robust. The Ce/Yb vs. Yb distribution (Fig. 15) suggests that PSAS melts form by 5–65% melting of an arc tholeiite source with a feldspar-free garnetiferous pyroxenite (eclogitic) residue. Extensive melting of aluminous source rocks where plagioclase is not a stable residual phase can account for the prominent positive Sr–Eu anomalies and steep L/HREE proﬁle of the model melts (Fig. 14a), and could help explain extensive plagioclase crystallisation upon ascent. This should not be taken to imply that only downthrust (subducted) volcanic facies are involved. The huge source volume needed to account for the size Grenville AMCG massifs (at least twice the observed volume of anorthosite and granite) suggests that the melting events may sample ‘whole-crust’ domains including cumulates and diﬀerentiated products also (Fig. 16b). High-Sr PSAS model melt can also be ﬁtted as a partial melt of the same Mt. Misery source (not shown), but with slightly more pyroxene and less amphibole in the residual mode. The high-Sr PSAS model melts, although generally very similar to the low-Sr model melts (Fig. 10b), have feldspar with much

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

97

Figure 14 Equilibrium melting models attempting to generate average low-Sr PSAS (a), and low-Sr ESAS and ES (b) from the average Mt. Misery arc tholeiite (Be´dard 1999), and from two Talkeetna arc cumulates (Greene et al. 2006). (c) Compares the average low-Sr ES melt to diﬀerent continental ﬂood basalt melt compositions. Proterozoic: North Nain Diabase and Seal Lake Group (Cadman et al. 1994, 1995a, 1995b); Michael gabbros (Emslie et al. 1997); Mt Lister intra-plutonic Dykes (Emslie et al. 1994). Mesozoic Anticosti dykes from Be´dard (1992).

lower An contents (Fig. 12), a feature that has yet to be fully explained. One possibility is that high-Sr andesine anorthosites might represent reworked low-Sr labradorite anorthosites (Buddington 1936; Berg 1969; Anderson & Morin 1969). Figure 14b shows variants of this Mt. Misery melting model that attempt to ﬁt average low-Sr ESAS melts. Apparently reasonable ﬁts are obtained only at low (ca. 5%) degrees of melting, and the residual assemblages are dominated by pyroxene with subordinate plagioclase and minor (1–2%) garnet. The need for such small amounts of garnet

suggests that it may not physically exist in the source, but may reﬂect natural variation in HREE Ds of clinopyroxene and amphibole. Small amounts of residual zirconrutile are also needed to produce satisfactory ﬁts for Zr and Ti. The LILE elements are over-estimated in the melting models, possibly indicating that the source had LILE-contents lower than those typifying the Mt. Misery basalts. The average high-Sr ESAS model can also be ﬁtted (not shown), but requires a diﬀerent residual mode, involving cpx/opx/plag/gt/ zircon/rutile=70/21/7/2/0·01/0·5, i.e. less clinopyroxene and

98

JEAN H. BEDARD

Figure 15 Ce/Yb vs. Yb (ppm) of model melts, compared to the Mt. Misery (S) equilibrium melting curves of Figure 14a and b. See text for discussion.

feldspar and more orthopyroxene. However, to account for the spread of the ESAS data in terms of a melting model (Fig. 15), melting degree is constrained to remain low, while the source volume had to change mode. Figure 14b also shows a melting model for average low-Sr ES melts. Reasonable ﬁts are obtained for small degree melting of the same Mt. Misery basalt source, but here the residue must be gabbronoritic, since abundant plagioclase is needed to buﬀer the Sr-abundance. The high-Sr ES melts could not be ﬁtted by partial melts of this source and appear to require a less depleted source rock (not shown). As for the ESAS models, the ES models intersect the target melt at low degrees of melting (e.g. Fig. 15), are very sensitive to the residual mode, and are not robust. The high compatible element abundances (Fig. 10a), the abundance of modal olivine in many ES rocks (Appendix Table A1 – Supplementary Material), and the resemblance of the ES trace element proﬁles to those of typical continental ﬂood basalts (Fig. 14c) and lamprophyres (Owens & Tomascak 2002; Fig. 11b) suggest that the ES melts may be mantle-derived. This would be consistent with the resemblance between Grenville Province ES model melts and Nain Plutonic Suite enriched maﬁc series model melts calculated by Be´dard (2001; Fig. 11b) that he interpreted as mantle-derived melts. If the ES model melts are mantle-derived, then extensive intracrustal fractional crystallisation is implied to account for the overall trace element enrichment. In this context, the high LILE and negative Nb–Ta–Ti anomalies could imply either extensive assimilation of crust (AFC), or derivation from a mantle previously enriched by arc processes. The constancy of the Ce/Yb ratios (Fig. 15) seems incompatible with the former, and favours derivation from subduction-modiﬁed mantle (Fig. 16b). Partial melting of delaminated crustal slabs may also contribute. If a dominant mantle derivation for ES series melts is accepted, then this suggests that ESAS and FAGS may represent mixtures of extensive high-pressure crustal PSAS melts with mantle-derived ES melts (Fig. 15). This mixing

scenario for ESAS and FAGS is more plausible than the melting models for these two suites.

7. Why are massif-type anorthosites solely a Proterozoic phenomenon? The evidence for a crustal arc source for AMCG magmas is consistent with phase equilibrium considerations and isotopic signatures (e.g. Demaiﬀe et al. 1986; Duchesne et al. 1989; Duchesne 1990; Van der Auwera et al. 1998; Longhi et al. 1999; Schiellerup et al. 2000), and (locally) inherited zircons (Scha¨rer et al. 1996). Heat could be provided either by thermal relaxation of tectonically thickened crust (Duchesne et al. 1999; Fig. 16a), or from underplating basaltic intrusions generated by post-tectonic delamination of the underlying mantle/slab (Corrigan & Hanmer 1997, ﬁg. 22b). The absence of AMCG complexes from the Phanerozoic is problematic in this regard, given that active margins abound in the postProterozoic record. Three possible explanations can be entertained. (1) The source was ephemeral, and was completely destroyed by ca. 900 my. This seems unlikely, since Archaean cratons persist in the geological record. Another model (2) posits that the anorthosites and associated rocks formed through melting of the base of a long-lived stable supercontinent (e.g. Hoﬀman 1989; Vigneresse 2005). However, the Grenvillean orogen was far from stable, and is in fact a collage of peri-continental and oceanic arc terranes assembled between ca. 1700 Ga and ca. 1100 Ga. Thermal blanketing scenarios appear to be inconsistent with the active-margin environment favoured by most of the geologists working in the Grenville Province (e.g. Rivers & Corrigan 2000; Hanmer et al. 2000; Gower & Krogh 2002). In addition, younger supercontinents (Rodinia and Pangea) are not characterised by anorthosites. It has also been suggested (3) that anorthositic massifs form when underthrust crustal rocks melt during post-orogenic

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS

99

Figure 16 (a) Cartoon illustrating the crustal tongue model of Duchesne et al. (1999) in a context of collisional tectonics, with island arc and continental blocks welded together and thickened. (1) High-K granites form above downthrust continental crust. (2) Lower-K granites and jotunitic melt parental to anorthosites form by fusion of juvenile simatic lower crust as it heats up when juxtaposed against hotter mantle (3). (b) Model proposed here, with scale of bodies adapted to Grenville Province AMCG Suite. After collisional orogenesis and crustal thickening, the eclogitised ‘oceanic’ slabs detach and sink, leading to large-scale mantle upwelling. This transfers abundant heat, which causes large-scale melting of lower crust. Low degrees of melting generate granitoids, whilst extensive melting generates melt parental to anorthosites, leaving behind abundant eclogitic residues (4) which must at least equal the volume of anorthosite+granite, and which may itself become unstable and delaminate. The inﬂux of heat also induces melting of subduction-modiﬁed mantle (5) and delaminated crust, so generating the associated maﬁc intrusions, some of which mix (6) with the anorthosites. Some of the granites may be residual melts expelled from fractionating anorthosite cumulates. Deeply-emplaced anorthosites may develop diapiric instabilities, while more shallowly-emplaced anorthosites (7) are conduit-fed (Royce & Park 2000).

thermal relaxation (Fig. 16; Duchesne et al. 1999; Corrigan & Hanmer 1997). In the context of this third model, the restriction of anorthosites to the Proterozoic is diﬃcult to explain. The absence of AMCG rocks from exhumed active margins (e.g. Greene et al. 2006; Jagoutz et al. 2007) implies that this is not just a sampling/erosional level problem. The only remaining possibility seems to be that the secular decrease in the activity of radioactive elements and mantle temperature (e.g. Kramers et al. 2001) has made it unlikely for large-scale remobilisation of the lower crust to occur in Phanerozoic orogens. If one accepts the post-tectonic model for AMCG genesis postulated above (hypothesis 3; Fig. 16), then the absence of

AMCG complexes from the Archaean is also problematic, given that the peripheral parts of many cratons are thought by many to be tectonic terrane collages (e.g. Percival et al. 2006; Smithies et al. 2007). The absence of pre-1 Ga high-pressure rocks (Stern 2005) and pre-2·5 Ga large-scale basaltic dyke swarms (Yale & Carpenter 1998); results of thermal modelling (Kramers et al. 1991); and the recognition of partial convective overturn events in craton cores (Chardon et al. 1996; Collins et al. 1998; Be´dard et al. 2003b), all imply that the preProterozoic crust was extremely weak, and was probably not capable of being thickened tectonically (Sandiford 1989; Bailey 1999; Be´dard 2006b). This provides a possible explanation for the absence of AMCG magmatism prior to 2·5 Ga.

100

JEAN H. BEDARD

8. Conclusions

10. Supplementary Material

Trace element inversion modelling of Grenvillean anorthosite massifs and associated maﬁc rocks emplaced from ca. 1·64 Ga to 1 Ga yield model melts with trace element proﬁles enriched in alkalis and light rare-earth elements (LREE) relative to normal mid-ocean ridge basalt (NMORB); commonly with negative Nb and Th anomalies. More than half of the models have steeply-fractionated trace element proﬁles with high Ce/Yb, positive Sr–Eu anomalies, and low abundances of HREE (heavy REE), and are named PSAS and FAGS (primitive steep anorthositic suite, and ﬂatter anorthositic-gabbroic suite) that cannot be linked through fractional crystallisation processes. However, PSAS and FAGS can be explained by extensive (5–65%), high-pressure (garnet in residue) partial melting of arc basaltic sources. The second most important model melt subtype generated from Grenvillean anorthosites is the enriched steep anorthositic suite (ESAS), which shares many of the PSAS traits, but at higher overall incompatible element abundances. ESAS melts could not be derived from PSAS parents by fractional crystallisation models. Partial melting models can generate ESAS melts from the same arc basalt source discussed above, but the models are not robust. A third important population of rocks yielded model melts with ﬂatter, more enriched proﬁles (enriched suite or ES). The incompatible element budgets of ES model melts can be modelled by lesser degrees of melting of the same arc source at low-pressures (plagioclase stable); but this is probably not consistent with the elevated abundances of compatible elements in these models, which suggests that the ES subtype represents partial melts of a mantle fertilised by arc magmatism. Mixing between ES and PSAS could provide a mechanism to generate ESAS- and FAGS-type melts. The active tectonic context now favoured for the Grenville Province appears to be inconsistent with plume or thermal insulation models. The heat source for melting could record either post-orogenic thermal relaxation, or basaltic underplating caused by delamination of a mantle root or subduction slab. This new model can explain the lithological associations, with the basaltic end-members forming from subductionmodiﬁed mantle, with anorthosites forming from the downthrust arc crust, and granitoids forming either from the less-depleted crustal segments, or being derived by fractionation of the anorthosite parental melt. In this context, preProterozoic anorthosites may be lacking, because prior to ca. 2·5 Ga, the lower crust was too weak to be thickened tectonically. It is notable that continent-scale dyke swarms are absent prior to 2·5 Ga. Post-Proterozoic anorthosites may be absent due to the secular decrease in radiogenic heating and cooling of the mantle, making extensive reactivation of thickened crust less likely.

The Appendix Table A1 is published as Supplementary Material with the online version of this paper. This is hosted by the Cambridge Journals Online Service and can be viewed at http://journals.cambridge.org/tre

9. Acknowledgements This is Geological Survey of Canada contribution #20080320. Jean-Franc¸ois Moyen and Gary Stevens are thanked for the invitation and means to present this paper at the Hutton Meeting. Louise Corriveau commented on an earlier version. Jean-Clair Duchesne and Lew Ashwal provided useful comments. Daniel Lamothe of the MRNFQ provided data extracts from the SIGE u OM database, and Claude He´bert (MRNFQ) provided unpublished data and valuable discussion. Pierre Brouillette (CGC-Que´bec) helped with the map and data projection.

11. References Amelin, Y. V., Larin, A. M. & Tuckerm, R. D. 1997. Chronology of multiphase emplacement of the Salmi rapakivi granite– anorthosite complex, Baltic shield: implications for magmatic evolution. Contributions to Mineralogy and Petrology 127, 353–68. Anderson, A. T. Jr. & Morin, M. 1969. Two types of massif anorthosites and their implications regarding the thermal history of the crust. In Isachsen, Y. W. (ed.) Origin of Anorthosite and Related Rocks. New York State Museum Science Service Memoir 18, 57–69. Anderson, J. L. 1983. Proterozoic anorogenic granite plutonism of North America. In Medaris, L. G., Byers, C. W., Mickelson, D. M. & Shanks, W. C. (eds) Proterozoic Geology. Geological Society of America Memoir 161, 133–54. Ashwal, L. D. 1993. Anorthosites. Berlin: Springer-Verlag, 422 pp. Bailey, R. C. 1999. Gravity-driven continental overﬂow and Archaean tectonics. Nature 398, 413–15. Be´dard, J. H. 1992. Jurassic high-Titanium quartz tholeiites from Anticosti Island, Que´bec. In Puﬀer, J. & Ragland P. C. (eds) Eastern North American Mesozoic Magmatism. Geological Society of America Special Paper SP268-09, 161–7. Be´dard, J. H. 1994. A procedure for calculating the equilibrium distribution of trace elements among the minerals of cumulate rocks, and the concentration of trace elements in the coexisting liquids. Chemical Geology 118, 143–53. Be´dard, J. H. 1999. Petrogenesis of boninites from the Betts Cove ophiolite, Newfoundland, Canada: Identiﬁcation of subducted source components. Journal of Petrology 40, 1853–89. Be´dard, J. H. 2001. Parental magmas of Nain Plutonic Suite anorthosites and maﬁc cumulates: A trace element modelling approach. Contributions to Mineralogy and Petrology 141, 747–71. Be´dard, J. H. 2005. Partitioning coeﬃcients between olivine and silicate melts. Lithos 83, 394–419. Be´dard, J. H. 2006a. Trace element partitioning in plagioclase feldspar. Geochimica et Cosmochimica Acta 70, 3717–42. Be´dard, J. H. 2006b. A catalytic delamination-driven model for coupled genesis of Archaean crust and sub-continental lithospheric mantle. Geochimica et Cosmochimica Acta 70, 1188–214. Be´dard, J. H. 2007. Trace element partitioning coeﬃcients between silicate melts and orthopyroxene: Parameterizations of D variations. Chemical Geology 244, 263–303. Be´dard, J. H., Page´, P. & Lissenberg, J. 2003a. Melt transfer mechanisms in the lower ophiolitic crust: Examples from the Bay of Islands, Thetford-Mines, Betts Cove and Annieopsquotch. EOS Transactions of the American Geophysical Union 84, p. F1540 [Abstract]. Be´dard, J. H., Brouillette, P., Madore L. & Berclaz A. 2003b. Archaean cratonization and deformation in the northern Superior Province, Canada: an evaluation of plate tectonic versus vertical tectonic models. Precambrian Research 127, 61–87. Be´dard, J. H., Leclerc, F., Harris, L. & Goulet, N. 2008. Intra-sill magmatic evolution in the Cummings Complex, Abitibi greenstone belt: Tholeiitic to calc-alkaline magmatism recorded in a subvolcanic conduit system. Lithos 111, 47–71. Berg, J. H. 1977. Regional geobarometry in the contact aureoles of the anorthositic Nain complex, Labrador. Journal of Petrology 18, 399–430. Berg, R. B. 1969. Petrology of anorthosites of the Bitterroot Range, Montana. In Isachsen, Y. W. (ed.) Origin of Anorthosite and Related Rocks. New York State Museum Science Service Memoir 18, 387–98. Blundy, J. D. & Wood, B. J. 1994. Prediction of crystal-melt partition coeﬃcients from elastic moduli. Nature 372, 452–4. Bowen, N. L. 1917. The problem of the anorthosites. Journal of Geology 25, 209–43. Bowen, N. L. 1922. The behaviour of inclusions in igneous magmas. Journal of Geology 30, 513–70. Brandriss, M. E. & Cawthorn, R. G. 1996. Formation of anorthosite and leucotonalite during magma hybridization in the Koperberg Suite of Namaqualand, South Africa. South African Journal of Geology 99, 135–51.

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS Bridgewater, D. D. & Windley, B. F. 1973. Anorthosites, postorogenic granites, acid volcanic rocks and crustal development in the North Atlantic Shield during the Mid-Proterozoic. In Lister, L. A. (ed.) Symposium on Granites, Gneisses and Related Rocks. Geological Society of South Africa Special Publication 3, 307–17. Buddington, A. F. 1936. Origin of anorthosite in the Adirondacks and in general. Transactions of the American Geophysical Union 17, 255–6. Cadman, A. C., Tarney, J., Baragar, W. R. A. & Wardle, R. J. 1994. Relationship between Proterozoic dykes and associated volcanic sequences: evidence from the Harp Swarm and Seal Lake Group, Labrador, Canada. Precambrian Research 68, 357–74. Cadman, A. C., Ryan, B., Royse, K. & Noble, S. R. 1995a. An investigation of some metamorphosed dykes of the Nain area: Part 3 – Geochemistry and structural relationships of dykes in the Nukasusutok Island area. In Current Research, Newfoundland Department of Mines and Energy, Geological Survey Branch Report 95-1, 1–14. St. John’s, Newfoundland: Newfoundland Department of Mines and Energy. Cadman, A. C., Tarney, J. & Baragar, W. R. A. 1995b. Nature of mantle source contributions and the role of contamination and in situ crystallisation in the petrogenesis of Proterozoic maﬁc dykes and ﬂood basalts Labrador. Contributions to Mineralogy and Petrology 122, 213–29. Chardon, D., Choukroune, P. & Jayananda, M. 1996. Strain patterns, decollement and incipient sagducted greenstone terrains in the Archaean Dharwar craton (South India). Journal of Structural Geology 18, 991–1004. Collins, W. J., Van Kranendonk, M. J. & Teyssier, C. 1998. Partial convective overturn of Archaean crust in the east Pilbara Craton, Western Australia: driving mechanisms and tectonic implications. Journal of Structural Geology 20, 1405–24. Corrigan, D. & Hanmer, S. 1997. Anorthosites and related granitoids in the Grenville orogen: A product of convective thinning of the lithosphere? Geology 25, 61–4. Corriveau, L., Perreault, S. & Davidson, A. 2007. Prospective metallogenic settings of the Grenville Province. In Goodfellow, W. D. (ed.) Mineral Deposits of Canada: A Synthesis of Major Deposittypes, District Metallogeny, the Evolution of Geological Provinces, and Exploration Methods. Geological Association of Canada, Special Publication 5, 819–48. Culshaw, N., Ketchum, J. & Barr, S. 2000. Structural evolution of the Makkovik Province, Labrador, Canada: tectonic processes during 200 Myr at a Paleoproterozoic active margin. Tectonics 19, 961–77. Davidson, A. 1998. Geological map of the Grenville Province, Canada and adjacent parts of the United States of America. Geological Survey of Canada Map 1947A, scale 1:2,000,000. Davidson, A. 2008. Late Paleoproterozoic to mid-Neoproterozoic history of northern Laurentia: An overview of central Rodinia. Precambrian Research 160, 5–22. Demaiﬀe, D., Weis, D., Michot, J. & Duchesne, J.-C. 1986. Isotopic constraints on the genesis of the Rogaland anorthositic suite (southwest Norway). Chemical Geology 57, 167–79. Dempster, T. J., Preston, R. J. & Bell, B. R. 1999. The origin of Proterozoic massif-type anorthosites: evidence from interactions between crustal xenoliths and basaltic magma. Journal of the Geological Society, London 156, 41–6. Dickin, A. P. & McNutt, R. H. 2007. The Central Metasedimentary Belt (Grenville Province) as a failed back-arc rift zone: Nd isotope evidence. Earth and Planetary Science Letters 259, 97–106. Duchesne, J.-C. 1984. Massif anorthosites: another partisan review. In Brown, W. L. (ed.) Feldspar and Feldspathoids. NATO Advanced Study Institute Series C137, 411–33. Dordrecht: Kluwer Academic. Duchesne, J.-C. 1990. Origin and evolution of monzonorites related to anorthosites. Schweizerische Mineralogische und Petrographische Mitteilungen Bulletin 70, 189–98. Duchesne, J.-C., Wilmart, E., Demaiﬀe, D. & Hertogen, J. 1989. Monzonorites from Rogaland (southwest Norway): a series of rocks coeval but not comagmatic with massif-type anorthosites. Precambrian Research 45, 111–28. Duchesne, J.-C., Lie´geois, J. P., Van der Auwera, J. & Longhi, J. 1999. The crustal tongue melting model and the origin of massif anorthosites. Terra Nova 11, 100–5. Dymek, R. F. & Owens, B. E. 2001. Petrogenesis of apatite-rich rocks (Nelsonites and oxide-apatite gabbronorites) associated with massif anorthosites. Economic Geology 96, 797–815. Emslie, R. F. 1978. Anorthosite massifs, rapakivi granites and late Proterozoic rifting of North America. Precambrian Research 7, 61–98.

101

Emslie, R. F. 1980. Geology and petrology of the Harp Lake Complex, Central Labrador: An example of Elsonian magmatism. Geological Survey of Canada Bulletin 293, 236 pp. Emslie, R. F. 1985. Proterozoic anorthosite massifs. In Tobi, A. C. & Touret, J. R. L. (eds) The Deep Crust in the North Atlantic Provinces. NATO Advanced Study Institute Series C158, 39–60. Dordrecht: D. Reidel. Emslie, R. F. 1991. Granitoids of rapakivi granite-anorthosite and related associations. Precambrian Research 51, 173–92. Emslie, R. F., Hamilton, M. A. & The´riault, R. J. 1994. Petrogenesis of a Mid-Proterozoic anorthosite–mangerite–charnockite–granite (AMCG) complex: Isotopic and chemical evidence from the Nain plutonic suite. Journal of Geology 102, 539–58. Emslie, R. F., Hamilton, M. A. & Gower, C. F. 1997. The Michael Gabbro and other Mesoproterozoic lithospheric probes in southern and central Labrador. Canadian Journal of Earth Sciences 34, 1566–80. Emslie, R. F. & Hegner, E. 1993. Reconnaissance isotopic geochemistry of anorthosite–mangerite–charnockite–granite (AMCG) complexes, Grenville Province, Canada. Chemical Geology 106, 279–98. Emslie, R. F. & Hunt, P. A. 1990. Ages and petrogenetic signiﬁcance of igneous mangerite–charnockite suites associated with massif anorthosites, Grenville Province. Journal of Geology 98, 213–31. Francis, D., Scowen, P., Panneton, G. & Doig, R. 2000. Contrasting Si-saturation in troctolite–anorthosite intrusions along the Manicouagan corridor of the Abitibi–Grenville transect. Canadian Journal of Earth Sciences 37, 271–89. Frith, R. A. & Currie, K. L. 1976. A model for the origin of the Lac St. Jean anorthosite massif. Canadian Journal of Earth Sciences 13, 389–99. Gobeil, A. He´bert, C., Clark, T., Beaumier, M. & Perreault, S. 2002. Ge´ologie de la re´gion du Lac De La Blache (22K03 et 22K04). Ministe`re des Ressources Naturelles du Que´bec, RG 2002-01. Gower, C. F. & Krogh, T. E. 2002. A U–Pb geochronological review of the Proterozoic history of the eastern Grenville Province. Canadian Journal of Earth Sciences 39, 795–829. Greene, A. R., DeBari, S. M., Kelemen, P. B., Blusztajn, J. & Clift, P. D. 2006. A detailed geochemical study of island arc crust: the Talkeetna Arc section, south-central Alaska. Journal of Petrology 47, 1051–93. Hanmer, S., Corrigan, D., Pehrsson, S. & Nadeau, L. 2000. SW Grenville Province, Canada: the case against post-1·4 Ga accretionary tectonics. Tectonophysics 319, 33–51. Hart, S. R. & Davis, K. E. 1978. Nickel partitioning between olivine and silicate melt. Earth and Planetary Science Letters 40, 203–19. He´bert, C., Cadieux, A. M. & van Breemen, O. 2005. Temporal evolution and nature of Ti–Fe–P mineralization in the anorthosite–mangerite–charnockite–granite (AMCG) suites of the south-central Grenville Province, Saguenay–Lac St. Jean area, Que´bec, Canada. Canadian Journal of Earth Sciences 42, 1865–80. He´bert, C., Cadieux, A. M. & van Breemen, O. 2009. Re´gion du re´servoir Pipmuacan (SNRC 22E): Synthe`se ge´ologique. Ministe`re des Ressources Naturelles et de la Faune, Que´bec. Report RG-2009-01+1/250,000 scale map. He´bert, C. & van Breemen, O. 2004a. Mesoproterozoic basement, the Lac St. Jean anorthosite suite and younger Grenvillian intrusions in the Saguenay region (Quebec): structural relationships and U–Pb geochronology. In Tollo, R. P., Corriveau, L., McLelland, J. & Bartholomew, M. (eds) Proterozoic tectonic evolution of the Grenville orogen in North America. Geological Society of America Memoir 197, 65–79. He´bert, C. & van Breemen, O. 2004b. Redeﬁnition of the Lac St. Jean anorthosite, central Grenville Province, Que´bec, based on compositional, structural, geochronological, and mineral deposit features. American Geophysical Union meeting, Montre´al. Abstract V53A-03. Higgins, M. D., Ider, M. & van Breemen, O. 2002. U–Pb ages of plutonism, wollastonite formation, and deformation in the central part of the Lac-Saint-Jean anorthosite suite. Canadian Journal of Earth Sciences 39, 1093–105. Hoﬀman, P. F. 1989. Speculations on Laurentia’s ﬁrst gigayear (2·0 to 1·0 Ga). Geology 17, 135–8. Icenhower, J. P., Dymek, R. F. & Weaver, B. L. 1998. Evidence for an enriched mantle source for jotunite (orthopyroxene monzodiorite) associated with the St. Urbain anorthosite, Que´bec. Lithos 42, 191–212. Jagoutz, O., Mu¨ntener, O., Ulmer, P., Pettke, T., Burg, J. P., Dawood, H. & Hussain, S. 2007. Petrology and Mineral Chemistry of Lower Crustal Intrusions: the Chilas Complex, Kohistan (NW Pakistan). Journal of Petrology 48, 1895–953.

102

JEAN H. BEDARD

Jochum, K. P., Hofmann, A. W., Ito, E. Seufert, H. M. & White, W. M. 1983. K, U and Th in mid-ocean ridge basalt glasses and heat production, K/U and K/Rb in the mantle. Nature 306, 431–6. James, D. T., Kamo, S. & Krogh, T. 2000. Preliminary U–Pb geochronological data from the Mealy Mountains terrane, Grenville Province, Southern Labrador. In Current Research 2000. Newfoundland Department of Mines and Energy Geological Survey, Report 2000-1, 169–78. St John’s, Newfoundland: Newfoundland Department of Mines and Energy. Kamo, S. L., Wasteneys, H., Gower, C. F. & Krogh, T. E. 1996. U–Pb geochronology of Labradorian and later events in the Grenville Province, eastern Labrador. Precambrian Research 80, 239–60. Ketchum, J. W. F., Culshaw, N. G. & Barr, S. M. 2002. Anatomy and orogenic history of a Paleoproterozoic accretionary belt: the Makkovik Province, Labrador, Canada. Canadian Journal of Earth Sciences 39, 711–30. Kramers, J. D., Kreissig, K. & Jones, M. Q. W. 2001. Crustal heat production and style of metamorphism: a comparison between two Archean high grade provinces in the Limpopo Belt, southern Africa. Precambrian Research 112, 149–63. Li, C. S., Lightfoot, P. C., Amelin, Y. & Naldrett, A. J. 2000. Contrasting petrological and geochemical relationships in the Voisey’s Bay and Mushuau intrusions, Labrador, Canada. Implications for ore genesis. Economic Geology 95, 771–99. Longhi, J., Fram, M. S., Van der Auwera, J. & Montieth, J. N. 1993. Pressure eﬀects, kinetics, and rheology of anorthositic and related magmas. American Mineralogist 78, 1016–30. Longhi, J., Van der Auwera, J., Fram, M. S. & Duchesne, J.-C. 1999. Some phase equilibrium constraints on the origin of Proterozoic (Massif) anorthosites and related rocks. Journal of Petrology 40, 339–62. Markl, G. & Frost, B. R. 1999. The origin of anorthosites and related rocks from the Lofoten Islands, northern Norway: II. Calculation of parental liquid compositions for anorthosites. Journal of Petrology 40, 61–77. Martignole, J. 1996. Tectonic setting of anorthositic complexes of the Grenville Province, Canada. In Demaiﬀe, D. (ed.) Petrology and Geochemistry of magmatic suites of rocks in the continental and oceanic crusts, 3–18. A volume dedicated to Professor Jean Michot, Universite´ Libre de Bruxelles. Tervuren, Belgium: Royal Museum for Central Africa. Martignole, J., Machado, N. & Nantel, S. 1993. Timing of intrusion and deformation of the Rivie`re-Pentecoˆte anorthosite (Grenville Province). Journal of Geology 101, 652–8. Martignole, J. & Schrijver, K. 1970. Tectonic setting and evolution of the Morin anorthosite, Grenville Province, Quebec. Geological Society of Finland Bulletin 42, 165–209. McLelland, J., Daly, J. S. & McLelland, J. M. 1996. The Grenville orogenic cycle (ca. 1350–1000 Ma): An Adirondack perspective. Tectonophysics 265, 1–28. Michot, P. 1965. Le magma plagioclasique. Geologische Rundschau 55, 956–76. Mitchell, J. N., Scoates, J. S. & Frost, C. D. 1995. High-Al gabbros in the Laramie anorthosite complex, Wyoming: Implications for the composition of melts parental to Proterozoic anorthosite. Contributions to Mineralogy and Petrology 119, 166–80. Mitchell, J. N., Scoates, J. S., Frost, C. D. & Kolker, A. 1996. The geochemical evolution of anorthosite residual magmas in the Laramie Anorthosite Complex, Wyoming. Journal of Petrology 37, 637–60. Morse, S. A. 1969. Layered intrusions and anorthosite genesis. In Isachsen, Y. W. (ed.) Origin of Anorthosite and Related Rocks. New York State Museum Science Service Memoir 18, 175–87. Morse, S. A. 1975. Plagioclase lamellae in hypersthene, Tikkoatokah Bay, Labrador. Earth and Planetary Science Letters 26, 331–6. Morse, S. A. 1982. A partisan review of Proterozoic anorthosites. American Mineralogist 67, 1087–100. Morse, S. A., Olson, K. E. & Hamilton, M. A. 1988. Massif anorthosite as a key to Proterozoic megatectonics. Geological Association of Canada Program with Abstracts 13, 87. Nantel, S. & Martignole, J. 1991. Le complexe anorthositique de Rivie`re-Pentecoˆte (Coˆte-Nord). Ministe`re de l’E u nergie et des Ressources, Que´bec Report MM89–02, 83 pp. Olson, K. E. 1992. The petrology and geochemistry of maﬁc igneous rocks in the anorthosite-bearing Adirondack Highlands, New York. Journal of Petrology 33, 471–502. Olson, K. E. & Morse, S. A. 1990. Regional Al–Fe maﬁc magmas associated with anorthosite-bearing terranes. Nature 344, 760–2. Owens, B. E., Dymek, R. F., Tucker, R., Brannon, J. C. & Podosek, F. A. 1984. Age and radiogenic isotopic composition of a late- to

post-tectonic anorthosite in the Grenville Province: the Labrieville Massif, Quebec. Lithos 31, 189–206. Owens, B. E. & Dymek, R. F. 1992. Fe–Ti–P-rich rocks and massif anorthosite: Problems of interpretation illustrated from the Labrieville and St-Urbain Plutons, Quebec. Canadian Mineralogist 30, 163–90. Owens, B. E. & Dymek, R. F. 2001. Petrogenesis of the Labrieville alkalic anorthosite massif, Grenville Province, Quebec. Journal of Petrology 42, 1519–46. Owens, B. E. & Dymek, R. F. 2005. Rediscovery of the Mattawa Anorthosite Massif, Grenville Province, Quebec. Canadian Journal of Earth Sciences 42, 1699–718. Owens, B. E. & Tomascak, P. B. 2002. Mesoproterozoic lamprophyres in the Labrieville Massif, Quebec: clues to the origin of alkalic anorthosites? Canadian Journal of Earth Sciences 39, 983–97. Pearce, J. A. & Parkinson, I. J. 1993. Trace element models for mantle melting: Application to volcanic arc petrogenesis: In Prichard, H. M., Alabaster, T., Harris, N. B.W. & Neary, C. R. (eds) Magmatic Processes and Plate Tectonics. Geological Society, London, Special Publication 76, 373–403. Bath, UK: The Geological Society Publishing House. Peck, W. H., Valley, J. W., Corriveau, L., Davidson, A., McLelland, J. & Farber, D. A. 2004. Oxygen-isotope constraints on terrane boundaries and origin of 1·18–1·13 Ga granitoids in the southern Grenville province. In Tollo, R. P., Corriveau, L., McLelland, J. & Bartholomew, M. J. (eds) Proterozoic tectonic evolution of the Grenville Orogen in North America. Geological Society of America Memoir 197, 163–82. Boulder, Colorado and Lawrence, Kansas: Geological Society of America and University of Kansas Press. Peck, W. H. & Valley, J. W. 2000. Large crustal input to high 18O anorthosite massifs of the southern Grenville Province: new evidence from the Morin Complex, Quebec. Contributions to Mineralogy and Petrology 139, 402–17. Percival, J. A., Sanborn-Barrie, M., Skulski, T., Stott, G. M., Helmstaedt, H. & White, D. J. 2006. Tectonic evolution of the western superior province from NATMAP and lithoprobe studies. Canadian Journal of Earth Sciences 43, 1085–117. Philpotts, A. R. 1976. Silicate liquid immiscibility: its probable extent and petrogenetic signiﬁcance. American Journal of Science 276, 1147–77. Philpotts, A. R. 1969. Parental magma of the anorthosite–mangerite suite. In Isachsen, Y. W. (ed.) Origin of Anorthosite and Related Rocks. New York State Museum Science Service Memoir 18, 207–12. Philpotts, A. R. 1966. Origin of anorthosite-mangerite rocks of southern Quebec. Journal of Petrology 7, 1–64. Rivers, T, 1997. Lithotectonic elements of the Grenville Province: review and tectonic implications. Precambrian Research 86, 117–54. Rivers, T. & Corrigan, D. 2000. Convergent margin on southeastern Laurentia during the Mesoproterozoic: tectonic implications. Canadian Journal of Earth Sciences 37, 359–83. Royce, K. R. & Park, R. G. 2000. Emplacement of the Nain anorthosite: diapiric versus conduit ascent. Canadian Journal of Earth Sciences 37, 1195–207. Sandiford, M. 1989. Horizontal structures in granulite terrains: A record of mountain building or mountain collapse? Geology 17, 449–52. Scha¨rer, U., Wilmart, E. & Duchesne, J.-C. 1996. The short duration and anorogenic character of anorthosite magmatism: U–Pb dating of the Rogaland complex, Norway. Earth and Planetary Science Letters 139, 335–50. Schiellerup, H., Lambert, R. D., Prestvik, T., Robins, B., McBride, J. & Larsen, R. B. 2000. Re–Os isotopic evidence for a lower crustal origin of massif-type anorthosites. Nature 405, 781–4. Schulz, K. J. & Cannon, W. F. 2007. The Penokean Orogeny in the Lake Superior region. Precambrian Research 157, 4–25. Scoates, J. S. & Chamberlain, K. R. 1997. Orogenic to post-orogenic origin for the 1·76 Ga Horse Creek anorthosite complex, Wyoming, USA. Journal of Geology 105, 331–43. Scoates, J. S. & Frost, C. D. 1996. A strontium and neodymium isotopic investigation of the Laramie anorthosites, Wyoming, USA: Implications for magma chamber processes and the evolution of magma conduits in Proterozoic anorthosites. Geochimica et Cosmochimica Acta 60, 95–107. Selbekk, R. S., Skjerlie, K. P. & Pedersen, R. B. 2000. Generation of anorthositic magma by H2O-ﬂuxed anatexis of silicaundersaturated gabbro: an example from the north Norwegian Caledonides. Geological Magazine 137, 609–21.

GRENVILLE PROVINCE ANORTHOSITE PARENTAL MAGMAS Simmons, E. C. & Hanson, G. N. 1978. Geochemistry and origin of massif-type anorthosites. Contributions to Mineralogy and Petrology 66, 119–35. Smithies, R. H., Van Kranendonk, M. J. & Champion, D. C. 2007. The Mesoarchean emergence of modern-style subduction. Gondwana Research 11, 50–68. Stern, R. J. 2005. Evidence from ophiolites, blueschists, and ultrahighpressure metamorphic terranes that the modern episode of subduction tectonics began in Neoproterozoic time. Geology 33, 557–60. Sun, S.-S. & McDonough, W. F. 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle compositions and processes. In Saunders, A. D. & Norry, M. J. (eds) Magmatism in the Ocean Basins. Geological Society, London, Special Publication 42, 313–45. Bath, UK: The Geological Society Publishing House. Taylor, S. R., Campbell, I. H., McCulloch, M. T. & McLennan, S. M. 1984. A lower crustal origin for massif-type anorthosites. Nature 311, 372–5. van Breemen, O. & Higgins, M. D. 1993. U–Pb zircon age of the southwest lobe of the Havre–Saint Pierre anorthosite complex, Grenville Province, Canada. Canadian Journal of Earth Sciences 30, 1453–7. van der Auwera, J., Longhi, J. & Duchesne, J.-C. 1998. A liquid line of descent of the jotunite (Hypersthene monzodiorite) suite. Journal of Petrology 39, 439–68.

103

Veksler, I. V., Dorfman, A. M., Borisov, A. A., Wirth, R. & Dingwell D. B. 2007. Liquid immiscibility and the evolution of basaltic magma. Journal of Petrology 48, 2187–210. Vigneresse, J. L. 2005. The speciﬁc case of the Mid-Proterozoic rapakivi granites and associated suite within the context of the Columbia supercontinent. Precambrian Research 137, 1–34. Wiebe, R. A. 1988. Structural and magmatic evolution of a magma chamber: The Newark Island Layered Intrusion, Nain, Labrador. Journal of Petrology 29, 383–411. Wiebe, R. A. 1990a. Dioritic rocks in the Nain Complex, Labrador. Schweizerische Mineralogische und Petrographische Mitteilungen Bulletin 70, 199–208. Wiebe, R. A. 1990b. Evidence for unusually feldspathic liquids in the Nain Complex, Labrador. American Mineralogist 75, 1–12. Wiebe, R. A. 1992. Proterozoic anorthosite complexes. In Condie, K. C. (ed.) Proterozoic Crustal Evolution, 215–61. Amsterdam: Elsevier. Woussen, G., Dimroth, E., Corriveau, L. & Archer, P. 1981. Crystallization and emplacement of the Lac St-Jean anorthosite massif (Quebec, Canada). Contributions to Mineralogy and Petrology 76, 343–50. Yale, L. B. & Carpenter, S. J. 1998. Large igneous provinces and giant dike swarms: proxies for supercontinent cyclicity and mantle convection. Earth and Planetary Science Letters 163, 109–22.

MS received 18 December 2007. Accepted for publication 22 July 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 105–115, 2010 (for 2009)

Granite formation: Stepwise accumulation of melt or connected networks? Paul D. Bons1, Jens K. Becker1, Marlina A. Elburg2 and Kristjan Urtson3 1

Mineralogy and Geodynamics, Institute for Geosciences, Eberhard Karls University Tu¨bingen, Wilhelmstr. 56, 72074 Tu¨bingen, Germany Email: [email protected]; [email protected]

2

Department of Geology and Soil Science, Ghent University, Krijgslaan 281 S8, 9000 Ghent, Belgium Email: [email protected]

3

Institute of Geology, Tallinn University of Technology, Ehitajate tee 5, 19086 Tallinn, Estonia Email: [email protected]

ABSTRACT: Several authors have proposed that granitic melt accumulation and transport from the source region occurs in networks of connected melt-ﬁlled veins and dykes. These models envisage the smallest leucosomes as ‘rivulets’ that connect to feed larger dykes that form the ‘rivers’ through which magma ascends through the sub-solidus crust. This paper critically reviews this ‘rivuletsfeeding-rivers’ model. It is argued that such melt-ﬁlled networks are unlikely to develop in nature, because melt ﬂows and accumulates well before a fully connected network can be established. In the alternative stepwise accumulation model, ﬂow and accumulation is transient in both space and time. Observations on migmatites at Port Navalo, France, that were used to support the existence of melt-ﬁlled networks are discussed and reinterpreted. In this interpretation, the structures in these migmatites are consistent with the collapse and draining of individual melt batches, supporting the stepwise accumulation model. KEY WORDS: accumulation

leucosomes, melt accumulation, melt networks, migmatite, Port Navalo, stepwise

The transport of crustally-derived magma, from mid/lowercrustal source rocks to (commonly) upper crustal emplacement levels, is one of the most important mass and heat transfer processes in the crust. The whole process, from initial melt formation, segregation, accumulation, to emplacement, spans an enormous range of length scales: about nine orders of magnitude from w10 m initial melt pockets to w10 km-scale plutons. The volume concentration factor is a staggering 1027. This large range in scales hampers the study of the complete process. One approach is to regard the whole process as a chain of distinct sub-processes, each operating on a typical length scale. This approach would separate the initial melt segregation into veins as one step, followed by accumulation of the melt in larger volumes, and then by ascent of the magma, typically in dykes, which ﬁnally leads to emplacement. Research over the past decades has had a tendency to investigate these four steps separately, possibly because their typical length scales necessitate diﬀerent research methods. The smallest scale is amenable to laboratory experiments (Rutter & Neumann 1995; Holtzman et al. 2003a, b; Walte et al. 2005; etc.), whereas the next step is best studied in the ﬁeld in migmatite terrains (e.g. Allibone & Norris 1992; Marchildon & Brown 2003; Sawyer 1996). As active dykes are diﬃcult to study in the ﬁeld, they have been mostly subject to geophysical modelling (e.g. Emerman & Marrett 1990; Rubin 1995a; Me´riaux et al. 1999; etc.). Pluton emplacement, ﬁnally, is again mostly studied in the ﬁeld (e.g. Zorpi et al. 1989; Paterson & Fowler 1993; Koukouvelas & Kokkolas 2003). The fact that diﬀerent research communities addressed diﬀerent steps in the process has left transitions from one step to another relatively neglected.

Another approach is to consider a single mechanism that covers the whole range from initial melt segregation to ﬁnal emplacement. Several mechanisms have been proposed: diapirism (e.g. Weinberg & Podladchikov 1994; Paterson & Vernon 1995), porous ﬂow (e.g. Jackson et al. 2003), fracture networks/dykes (e.g. Weinberg & Searle 1998; Nicolas & Jackson 1982; Weinberg 1999; Olson et al. 2004) and step-wise accumulation (e.g. Maaløe 1987; Bons & van Milligen 2001; Bons et al. 2001a, b, 2004). Diapirism mostly ‘avoids’ the problem of segregation and accumulation by envisaging wholesale mobilisation of partially molten rock. However, most workers nowadays regard diapirism as a non-viable mechanism for crustal pluton formation (Clemens & Mawer 1992; Vigneresse 2004). Most modelling of dykes has not considered the question of how the dykes are supplied with magma, as it is normally assumed that there is either a constant magma pressure or ﬂux at the base (Emerman et al. 1986; Spence et al. 1987; Rubin 1995a; Me´riaux & Jaupart 1998). Weinberg (1999) has been among the few authors to address this question. He and others proposed the formation of fracture networks where small fractures feed larger ones. This ‘rivulets-feeding-rivers’ or ‘rooted vein network’ model has been applied to both the mantle (Nicolas 1986; Hart 1993; Maaløe 2003) and the crust (Brown & Solar 1998; Petford & Koenders 1998; Weinberg 1999). The concept is appealing in that a single process can span the whole process from large dykes that feed plutons down to the smallest fractures that tap melt from between grains. However, the present paper will discuss several problems with the rivulets-feeding-rivers model. The paper will address ﬁeld evidence and theoretical problems of the model,

2009 The Royal Society of Edinburgh. doi:10.1017/S175569100901603X

106

PAUL D. BONS ET AL.

Figure 1 Schematic overview of the transport steps in the RFR model. A developing granite pluton is fed by a feeder dyke, which in turn taps a hierarchical network of melt-ﬁlled veins. This network spans the scale range from the feeder dyke down to the smallest veins that are fed by porous ﬂow and/or diﬀusion with melt from grain boundaries (inset).

and it will be argued that discontinuous connectivity of melt-ﬁlled fractures in a stepwise accumulation model is a preferred model. The discussion will be limited to the formation and transport of purely crustally-derived melts, as an end member of granite petrogenesis, even though it is recognised that many granites contain at least some mantle input (Elburg 1996; Soesoo & Nicholls 1999). The word ‘leucosome’ will be used for the ﬁrst stage of melt vein that accumulates although it is recognised that some leucosomes, as found in outcrop, do not represent melt compositions, but cumulates (Sawyer 1987; Ellis & Obata 1992; Brown et al. 1999; Solar & Brown 2001; Johannes et al. 2003). Furthermore, the term ‘melt’ will be used for the material inside leucosomes, even though it may contain solid material, such as residue or newly formed crystals.

1. Fracture networks The rivulets-feeding-rivers (RFR) model can be summed up by citing Weinberg (1999): ‘In order to produce a transporting dyke, the source must evolve to a stage of maturity in which an extensive tributary dyke network is capable of maintaining the high flow rates in the transporting dyke’. In 1994, Brown had already proposed essentially the same model: ‘Melt is extracted from the accumulation networks by porous flow down gradients in pressure to ductile opening mode fractures and shear zones, where ascent is buoyancy driven. Extraction occurs at some critical combination of melt fraction and distribution, most probably as the developing accumulation network reaches the percolation threshold’. Both authors envisaged a connected network of tributary melt-ﬁlled fractures that must reach a state of maturity to enable ﬂow to occur from the source towards one or more dykes that transport the magma through the crust. To address the viability of this model, one must consider the following questions: + Can such a mature, extensive tributary dyke network form? + Is the model feasible, not only considering length scales, but also time scales? + What is the ﬁeld evidence for connecting networks? These questions will be addressed in the following sections.

1.1. Can extensive melt-filled fracture networks exist? In the RFR model, the process from melting at grain boundaries to formation of a pluton is generally divided in ﬁve steps (e.g. Brown 2004) (Fig. 1): 1. melting of the source rock,

2. transport of melt along grain boundaries towards melt-ﬁlled veins (leucosomes), 3. transport of the melt towards a large conduit (feeder dyke), 4. transport of magma across subsolidus crust through the conduit, 5. emplacement of the magma. Some authors combine steps 3 and 4 into one, eﬀectively regarding the ﬁnal conduit, or feeder dyke, as the largest melt-ﬁlled vein in a single system of connecting veins (e.g. Weinberg 1999; Harris et al. 2000; Vigneresse 2004). The time and length scale of step 4 is probably the one constrained best, because transport through the feeder dyke must be fast enough to inhibit freezing of the ascending magma in the subsolidus crust. Critical dyke width and magma ascent rate depend on buoyancy and viscosity of the magma. Estimated minimum dyke widths are in the order of metres and ﬂow rates up to cm/s, meaning that a pluton may be ﬁlled in a matter of tens to thousands of years in a single event (Clemens 1998). For the very eﬃcient dyke transport step to work, enough melt must be able to drain into the base of the feeder dyke (Weinberg 1999). This is envisaged to occur by the formation of a percolating network of melt-ﬁlled fractures. This network must be able to support a suﬃcient ﬂux to keep the magma in the ﬁnal conduit ﬂowing without freezing and clogging the system. To achieve this, and to provide a link from the smallest veinlets up to the largest dyke, a hierarchical or self-organised network is envisaged (Weinberg 1999; Vanderhaeghe 2001; Brown 2004; Moyen et al. 2003; Maaløe 2003) (Fig. 1). Many small veins feed fewer veins that are one order in size larger. These connect to even fewer of the next order up, and so on. This idea is supported by Tanner (1999), who suggested that melt-vein networks have a fractal structure that is similar to a Menger Sponge. The present paper will not question whether a melt-ﬁlled network can feed a feeder dyke (which it can if the decrease in number of higher-order dykes is balanced by their increased width), but whether such a network is actually likely to form in nature. A critical point is that the network must ﬁrst develop before it can feed a pluton. The ﬁrst step in forming a pluton is the formation of melt, which is controlled by heat diﬀusion or decompression rates, depending on the cause of melting. The time scale of melt formation is thus controlled by the duration of tectono-metamorphic events, which is in the order of 1–10 million years. This time scale is much larger than the time needed to ﬁll a pluton (Petford et al. 2000 and references therein). One possibility is that the heating events themselves are caused by underplating/intraplating events of hot basaltic magma into the crust, causing a relatively short burst of

STEPWISE ACCUMULATION OF MELT OR CONNECTED NETWORKS?

107

melting. Petford & Koenders (1998) suggest a time frame in the order of 105 years. If melting is not in such short bursts, magma-extraction events are separated by prolonged periods in which melt is formed, but no melt escapes from the source. In the RFR model, the networks form during these intervals, implying that the networks must be sustainable, without signiﬁcant ﬂow, for a prolonged period of time.

1.2. The percolation threshold Several authors have used the concept of a percolation threshold to determine the melt fraction at which melt can ﬂow (see e.g. citation above of Brown 1994; Vigneresse et al. 1996; Weinberg 1999; Vigneresse & Burg 2000). In this concept, the percolation threshold (PT) or ‘melt escape threshold’ (Vigneresse et al. 1996) is the fraction of melt at which the melt-ﬁlled conduits (pores, veins) link up to form a percolating network, through which melt can ﬂow over large distances (relative to the size of single pores or veins). Percolation thresholds or critical melt fractions of around 10–20% are usually quoted (Vigneresse et al. 1996). The actual value of the PT depends on several factors, such as shape and orientation of the conduits. Randomly oriented fractures have the highest chance of intersecting, and hence the lowest PT. However, there is abundant ﬁeld evidence that melt-ﬁlled veins are generally aligned parallel to structures and foliations, or are oriented by the stress ﬁeld, thus decreasing their chance of intersecting and increasing the PT. Deformation, however, may also aid in creating connectivity and probably lowers the PT. Some authors assume that there is no ﬂow below the PT or critical melt percentage (e.g. Brown 1994; Weinberg 1999; Vigneresse 2007). Other authors make no speciﬁc statement on how a percolating network develops, and how much transport takes place during this stage, as their modelling is based on the assumption that the network has developed (Hart 1993; Maaløe 2003). A fully percolating network, of whatever geometry (rooted vein network or other), would develop by the formation of more and more melt-ﬁlled veins as the melt fraction increases. Contrary to the statement of Weinberg (1999) that ‘the permeability is zero below a critical value’ (the PT), the permeability of the system below the PT is non-zero and increases as more melt produces more melt-ﬁlled veins that have an increasing chance of intersecting (Petford & Koenders 1998; Petford et al. 2000). This means that melt can start to ﬂow and change the developing network before it reaches the PT. Flow from one vein into another may lead to eﬀective closure of the ﬁrst vein, thus reducing the number of melt-ﬁlled veins and, hence, the permeability. To achieve a fully percolating network, connectivity creation by melting must outpace connectivity reduction by ﬂow. A rough estimate of the rate of connectivity, and hence permeability destruction can be made by considering a limited cluster of connected veins. One driving force for ﬂow within a limited cluster of connected veins is the buoyancy of the melt. This will drive melt from a lower vein into a vein higher up, to which it is connected. If the slow buoyancy-driven ﬂow can be shown to be eﬃcient enough to inhibit the development of a fully percolating network, there is no need to consider the additional eﬀect of deformation. The rate at which melt would ﬂow from one vein into a second one is limited by the ductile ﬂow of the matrix, which must accommodate the closure of the ﬁrst and opening of the second vein. A simple geometry of connected fractures may serve to estimate the time-scale of network collapse by ductile ﬂow of the solid matrix. Consider a volume of rock, with dimensions 2LLL, in the source zone. There is a thin horizontal sill at the base of the box, as depicted in Figure 2a. The sill has a horizontal extent

Figure 2 (a) Idealised simple fracture network, consisting of horizontal fractures at the top and bottom of the box that are connected by a vertical fracture. The bottom fracture is ﬁlled with melt and has a maximum width W, which is much smaller than the size of the box, deﬁned by L. (b) After time t, all melt is transferred to the top fracture and the bottom fracture is closed. Transfer rate is controlled by the shear rate of the solid matrix. (c) Duration (t) to transfer all melt from bottom to top fracture as a function of size of the network, for three diﬀerent combinations of width of the melt-ﬁlled fracture (W) and viscosity () of the solid matrix. A tall network (tens to hundreds of meters) cannot be maintained for more than a few (hundreds of) years.

of 2LL and a maximum thickness W<
(1)

For this idealised geometry, the deformation (induced by the ﬂow of the melt) of the matrix can be approximated by simple shear in the solid matrix with a vertical shear plane. The duration (t) to full closure of the bottom sill is determined by the simple shear rate simple shear rate ˙: ˙ ⫽

W Lt

(2)

The work done (E) to close the bottom sill is a function of the shear rate, the shear stress (), the volume of the deforming material and the duration: Ey =˙L3t

(3)

Inserting equation (2) into (3), and relating the strain rate to the shear by the viscosity (=˙) then gives the work done by shearing of the matrix: 2

2 3

W L t

Ey⫽˙ L t⫽

(4)

Finally, the time needed to transfer the melt from the bottom sill to the top sill can be found by equating Ep and E, and rearranging:

108

PAUL D. BONS ET AL. 2

3

W W L 5t⫽ 2 t gL

Ep⫽E5gWL ⫽

(5)

Clearly the time needed for the transfer increases with increasing viscosity of the matrix and width of the fracture (amount of melt). The time decreases with increasing buoyancy of the melt (, here taken at 300 kg/m3) and vertical extent of the connected fractures (L). Considering that the solid matrix is in or close to the source region of magma, it can be regarded as relatively weak. The matrix viscosity is conservatively estimated at =1016–1018 Pa·s. Figure 2c shows that the time to close the local network is in the order of years to hundreds of years for a vertical extent in the order of tens to hundreds of meters. This means that it is diﬃcult to build or maintain a tall network of connected melt-ﬁlled fractures over a prolonged period of time (>> thousands of years). One could argue that a fracture cannot fully close because an inﬁnite time is needed to squeeze out a viscous ﬂuid from between two parallel (elastic) plates. However, in nature, especially under the high metamorphic conditions applicable here, ductile ﬂow, dissolution– precipitation processes and dynamic recrystallisation would enable fractures to eﬀectively close.

1.3. Hydraulic fracturing The fully percolating network invoked by the RFR model would have a large vertical and horizontal extent. The vertical extent would be approximately the height of the partially molten zone, estimated at R1 km. This implies that, before the PT, developing clusters should reach vertical extents of at least hundreds of metres. Such clusters would not be stable. Weertman (1971) ﬁrst suggested that vertical fractures that are ﬁlled with a buoyant liquid have a maximum vertical extent. Above the critical height, the fracture becomes unstable and starts to propagate upwards (Takada 1990; Secor & Pollard 1975). The critical height depends on several parameters, such as the density diﬀerence between melt and matrix and the fracture toughness (or fracture energy: Rubin 1995b). It can be estimated to be about R100 m (Secor & Pollard 1975), which means that a developing fracture network would become unstable well before it has reached its full vertical extent. Instead, developing clusters would start to propagate upwards, draining the melt from below and destroying connectivity.

2. Stepwise accumulation A feasible model for the extraction of magma from the source regions in the lower to middle crust must reconcile the relative long duration of melt formation with occasional short ‘bursts’ of draining and transport of magma to pluton emplacement levels. Field evidence suggests that initial melt tends to drain into melt-ﬁlled veins (leucosomes) from the 1–10 cm scale upwards. Veins generally develop parallel to existing foliations (layering, cleavage) and in dilatant sites, such as boudin necks, as has been extensively described by e.g. Brown et al. (1995), Brown (2005a), Vanderhaeghe (2001), etc. There seems to be general consensus that after initial segregation of melt into veins (by porous ﬂow or diﬀusion), most transport and accumulation takes place through these veins. With increasing number and volume of the veins, transport and accumulation of melt can commence, well below the PT. Vertical ﬂow can be driven by buoyancy of the melt, ﬁlling the uppermost veins in a locally connected vein cluster. The buoyancy of the melt may also induce upwards propagation of fractures, but only if a critical length of >100 m is exceeded. However, deformation may induce fracture propa-

gation of much shorter veins, because the normal stress gradients acting on a vein can be much larger (up to about one MPa/m) than those resulting from the density diﬀerence between melt and matrix (%0·003 MPa/m). The eﬀect of these processes is to form ever-larger volumes of melt. These ‘batches’ are mostly isolated, because any connectivity would lead to melt ﬂow and accumulation. Connectivity is therefore transient, with the duration of local connectivity decreasing with increasing size of the connected system (Fig. 2). The process of transient ﬂow and accumulation without ever reaching full percolation is illustrated by the highly simpliﬁed (‘toy’) experiment of Bons & van Milligen (2001). In the experiment, yeast is added to loose sand that is saturated with sugar water. Fermentation of the dissolved sugar produces alcohol and CO2 gas. As the amount of gas increases, horizontal gas-ﬁlled hydrofractures appear (Fig. 3). These hydrofractures appear, grow, and then collapse again by draining into other hydrofractures higher up in the tank. There is never a fully percolating network of gas-ﬁlled fractures, yet there is an upward ﬂux of gas. Upward ﬂow takes place by short transient events when local connectivity is brieﬂy established, and subsequently destroyed as the gas ﬂows from one hydrofracture to another. As stated above, the process of stepwise accumulation and ascent (Bons et al. 2001b) leads to the formation of ever-larger volumes of melt (or gas in the experiment). This eﬀect is enhanced by the occurrence of transport avalanches, which are also observed in the experiment. The collapse of one hydrofracture may trigger a chain reaction of interactions between other hydrofractures. Such a chain reaction can suddenly drain a large fraction of the gas within the tank. Numerical modelling by Bons et al. (2004) showed that power–law size distributions of accumulated volumes result from this process. A power–law distribution of the volumes of gas-ﬁlled experimental hydrofractures was also reported by Urtson & Soesoo (2007). Field evidence for power–law size distributions will be discussed below.

3. Field observations Both the RFR and the stepwise accumulation (SA) models should be testable in the ﬁeld. Evidence for connecting melt networks is expected for the RFR model, but not for the SA model. The SA model predicts that (1) former melt-ﬁlled veins have a power–law size distribution, and (2) structures should be present that represent collapsed former melt volumes. Migmatite outcrops near Port Navalo on the South Brittany coast (France) were chosen to investigate this, because these outcrops have been studied in detail by Brown (1983), Marchildon & Brown (2003) and Johnson & Brown (2004). The reader is referred to these papers for a thorough description of the local geology, which is only brieﬂy recapitulated here. The study area is located in the Southern Brittany Metamorphic (or Migmatite) Belt (SBMB), which is part of the Variscan Armorican arc. The area was deformed and metamorphosed during the Paleozoic Variscan orogeny, reﬂecting interaction (subduction, collision and intra-continental deformation, followed by extension and exhumation) between Laurasia and Gondwana (Matte 2001). The area is located to the southwest of the South Armorica shear zone, which separates South from Central Armorica. It is separated from lower-grade rocks to the south by a late-orogenic extensional detachment (Gapais et al. 1993). The study area is dominated by metapelites, but amphibolites are present too. Both lithologies display signs of partial melting, but migmatisation is better

STEPWISE ACCUMULATION OF MELT OR CONNECTED NETWORKS?

109

Figure 3 Five stages (six-minute intervals) in an experiment where CO2 gas is produced by fermentation in sand that is saturated with sugar water (Bons & van Milligen 2001). The buoyant gas props open hydrofractures (black arrows) that subsequently become unstable and drain (white arrows) into other hydrofractures higher up in the system. Full percolation of the gas phase is never reached, yet there is a non-zero upwards ﬂux of gas. First image is after about t0 =2 hours after onset of fermentation. Width of view about 15 cm. See Bons & van Milligen (2001) for a detailed description of the experiment.

developed in the metapelites. Leucosomes and granitic dykes consist of plagioclase, quartz, perthitic K-feldspar, myrmekitic intergrowths and biotite with red-brown to straw-yellow pleochroism. Zircon and apatite occur as minor accessory phases. The melanosomes consist of the same phases plus cordierite or garnet, and contain higher proportions of biotite and accessory phases. Peak metamorphic conditions associated with a ﬁrst phase of melting have been estimated at 8 kbar and 800(C, whilst a second phase of melting resulted from nearisothermal decompression (at 700(C) to w4 kbar (Johnson & Brown 2004). Ar–Ar cooling ages for hornblende and muscovite from the area are 303–298 Ma and 306–305 Ma

respectively (Brown & Dallmeyer 1996). Three deformation phases have been recognised in the area (Marchildon & Brown 2003).

3.1. Leucosomes and granite dykes The subhorizontal wave platforms along the coast provide an ideal perpendicular section through the subvertical structures in the stromatic migmatites that contain former melt. There are three types of such structures: + Leucosomes parallel to bedding and bedding-parallel foliation, together S01, contain most of the former melt (up to

110

PAUL D. BONS ET AL.

Figure 4 Images from the subhorizontal migmatite outcrops at Port Navalo, France: (a) Coarse-grained, bedding-parallel leucosomes are cut by a younger aplite dyke. There is clear petrographic discontinuity between leucosomes and aplite dyke. Some of the leucosomes on the left show isoclinal folding. Scale bar=50 cm; (b) Second example of an aplite dyke cutting coarser-grained, bedding-parallel leucosomes, again with a clear petrographic discontinuity; (c) Cross-cutting leucosome in boudin-like structure; (d) Cross-cutting leucosome with dextral shear, displaying both reverse and normal drag folds of the adjacent foliation. Scale bar=20 cm; (e) Boudin-like structure in amphibolite layer. The strong bending of the layering can only be explained by collapse of a melt volume by melt loss (Brown 2005b). Scale bar=5 cm.

about 40%). Leucosomes are folded together with the S01 foliation. Most folds are open folds, attributed to D3 (Marchildon & Brown 2003). Isoclinally folded leucosomes with an axial plane parallel to S01 can sometimes be observed (Fig. 4a). + Discordant leucosomes have a similar composition and internal texture as the bedding-parallel leucosomes (Fig. 4c–d). These leucosomes typically form boudin- and shear band-like structures, where the S01 foliation is deﬂected. Boudins show typical ﬁsh-mouth structures where the foliation pinches in towards the boudin neck. The foliation bends into shear band-like leucosomes (Fig. 4d), whereby the drag direction can be both normal (synthetic with sense of shear) and reverse (antithetic with respect to sense of shear) (Grasemann & Stu¨we 2001; Exner et al. 2004). + Granitic and aplitic dykes cut all the previously mentioned leucosomes (Fig. 4a–b). The dykes have a ﬁner grain size

than the leucosomes. The dykes have widths of tens of centimetres to metres, are steeply dipping, but can have variable strikes, which means they sometimes intersect. The dykes are mostly straight and appear not or only little aﬀected by the latest folding. These dykes therefore formed late in the history of the migmatites, and must post-date at least part of the leucosome formation. Marchildon & Brown (2003) and Brown (2004, 2005a) argue that the dykes and the layer-parallel leucosomes show ‘petrographic continuity’, implying they both contained the same melt at the same time (e.g. ﬁgure 3D in Marchildon & Brown 2003). They also argued that the intersecting dykes show no clear crosscutting relationships (ﬁgure 6A in Marchildon & Brown 2003), again suggesting that the intersecting dykes were all ﬁlled with melt at the same time. These interpretations lead to a model whereby melt ﬂows from bedding-parallel

STEPWISE ACCUMULATION OF MELT OR CONNECTED NETWORKS?

111

Figure 5 (a) Intersecting granite dykes in subhorizontal outcrop at Port Navalo. Dykes are steeply dipping (same intersection as shown in ﬁgure 6A of Marchildon & Brown 2003). Bag for scale in upper-right corner. (b) Close observation shows that the intersection is of three dykes of diﬀerent generations, each cutting older dykes and the wavy stromatic layering.

leucosomes into dilatant discordant leucosomes (both shear bands and boudin necks), which subsequently feed into a network of dykes. Flow would have been perpendicular to the current horizontal outcrops, focussed in tubular intersections of dykes. Close observation of the outcrops leads the present authors to disagree with the above interpretation. Dykes do clearly crosscut the leucosomes and show a petrographic discontinuity, with the dykes having a diﬀerent grain size and colour (Fig. 4a–b). Furthermore, dykes cut isoclinally folded leucosomes, whereas they themselves are not folded (Fig. 4a). It is therefore concluded that the dykes and the leucosomes (both bedding parallel and discordant) belong to diﬀerent generations and were not ﬁlled with melt at the same time. This two-stage melting scenario is supported by the petrological work of Johnson & Brown (2004). The present authors, however, see no indication that the rocks remained molten throughout these two melting events. There is little variation in composition and texture of the dykes, which makes it diﬃcult to distinguish generations where dykes cut each other, as in Figure 5a. Yet, even there, subtle textural diﬀerences and the geometry of the dykes show that diﬀerent generations of dykes intersected each other (Fig. 5b). The straight and sharp boundaries between the diﬀerent generations indicate that the older dyke was already solidiﬁed when cut by a younger one. Therefore, no evidence is seen for a melt-ﬁlled network of dykes at the Port Navalo migmatite outcrops.

Figure 6 Cartoon showing the formation of boudins and shear bands by removal of melt. (a) Situation before melt loss, where melt is black. White arrows show convergence direction of the wall rock when the melt pockets collapse. (b) Situation after melt removal, showing a typical ﬁsh mouth boudin (right) and shear bands with reverse drag folds (left). Notice that there is no bulk extension.

3.2. Interpretation of the discordant leucosomes Discordant leucosomes, both shear bands and boudin necks, are commonly interpreted as dilational structures (e.g. Hollister & Crawford 1986; Oliver & Barr 1997; Kisters et al. 1998; Brown 2004). For boudin necks, this interpretation seems obvious at ﬁrst sight, as the separation of boudins creates a pressure gradient that could suck in melt (Allibone & Norris 1992). Depending on orientation, shear fractures may also induce pressure gradients that lead the melt to ﬂow towards them (Sleep 1988). The concentration of melt in shear zones is furthermore observed in grain-scale experiments (e.g. Rosenberg & Handy 2001; Holtzmann 2003a, b; Walte et al. 2005). Despite the apparent plausibility of the interpretation that the discordant veins represent dilatant structures, the possibility should also be considered that these structures are in fact the opposite: namely contraction structures. When melt drains from an accumulated volume of melt, the surrounding rocks converge. Figure 6 shows that, depending on the original shape of the melt volume, a variety of structures may form, ranging from shear band-type leucosomes to

apparent boudins (Bons 1999; Kriegsman 2001). Shrinkage of equidimensional melt volumes produces boudin-like structures as the surrounding foliation converges from all sides. This produces the typical ﬁsh-mouth boudin necks, so commonly observed in migmatites, and which Kriegsman (2001) used to quantify the amount of melt loss. Shrinkage of lenticular veins that are oblique to the foliation produces shear band-like leucosomes, which are typically ﬂanked by reverse drag folds (bending of the foliation opposite to the apparent sense of shear, Grasemann & Stu¨we 2001). The deﬂection of the foliation is, however, normal (synthetic with the apparent oﬀset) at the tips of the collapsing veins. The combination of both reverse and normal drag at the centre and tips of a leucosome, respectively, is commonly observed in migmatites (Fig. 4d). Experiments by Druguet & Carreras (2006) show that deformation enhances this eﬀect. The collapse of irregular melt volumes leads to more complex structures with the remaining melt left in irregular, spider-like leucosomes (Fig. 4e). It is of interest to note that Brown (2005b) actually

112

PAUL D. BONS ET AL.

Figure 7 (a) Width distribution of leucosomes in migmatites near Port Navalo in a log–log graph of number (N>W) of leucosomes wider than a width W against that width in mm. Our data (ﬁlled circles & squares) are from two parallel proﬁles that were logged perpendicular to the leucosomes. The leucosome fraction was 31%. Leucosomes wider than about 5 mm follow a power–law distribution. Note the similarity in distribution with the data from Brown (2005b), extracted from his ﬁgure 8 (open circles). It is however not certain that his measurements are on the same generation dykes as the leucosomes analysed by the present authors. (b) Photograph of the 3 m-long section. Note that the crosscutting aplite dyke was not included in the log.

interpreted this structure as indicating melt loss, but the other discordant veins at the same location as dilatant sites. Is it possible to distinguish dilatant from collapse structures? There are unambiguous cases, such as is shown in Figure 4e. However, in many other cases both interpretations seem to be permissible. Shearing along a fracture or vein in a ductile matrix leads to the formation of both normal and reverse drag folds by geometric necessity (Gomez-Rivas et al. 2007). Similarly, extensional boudinage leads to pinching in of the foliation. These structures as such are therefore not suﬃcient to distinguish dilation or contraction. What needs to be considered is whether the amount of deﬂection of layers is consistent with either model. Unfortunately, the formation of collapse structures has not yet been investigated in any detail. The strong deﬂections for relatively short shear band-type leucosomes (Fig. 4d), as well as the presence of unambiguous collapse structures (Fig. 4e) lead the present authors to favour melt loss and contraction as the origin of these structures. The fact that leucosomes often have a cumulate composition supports the model of melt loss, whereby solid cumulate material would preferentially remain behind (Ellis & Obata 1992; Brown et al. 1999; Johannes et al. 2003).

3.3. Power–law distributions Having established that melt-loss structures are found in migmatites, we now turn to the prediction that melt volumes should exhibit power–law volume distributions. In the ﬁeld it is virtually impossible to measure volume distributions of leucosomes, as one normally only has one-dimensional (drill core) or two-dimensional (surface) outcrops. However, if leucosomes have a power–law volume distribution, they are also

expected to have a power–law width distribution. This was indeed observed in migmatites from South Finland (outcrop) and the Estonian basement (drill core) (Bons et al. 2004; Soesoo et al. 2004). Power–law width distributions were also reported by Kruhl (1994) for a range of scales from small leucosomes to dykes. A power-law distribution of vein widths can be described by the following equation, where N>W is the number of veins wider than W, and n is an exponent: N >WfW n

(6)

Brown (2005a) reports a power–law exponent of n=1·11 for dykes wider than about 10 cm at Port Navalo. Marchildon & Brown (2003) measured width distributions of thinner leucosomes at Port Navalo, and conclude that these do not follow a power law. Their conclusion is, however, based on all recorded leucosomes, down to a width of one mm. Such widths are in the order of the grain size in the rock, which means it is diﬃcult to record them all. This is one of the truncation eﬀects from which this technique invariably suﬀers, and which are most pronounced at the lower end of the measured spectrum (Bonnet et al. 2001). The present authors analysed two proﬁles (3 m and 4 m long, respectively), in the same migmatites at Port Navalo, and observed a power–law width distribution (n=1·3) for leucosomes wider than 5 mm (Fig. 7). Both these data (widths 5 mm to w100 mm) and those of veins/dykes wider than 100 mm of Brown (2005a) each span less than two orders of magnitude, which is a small range to conﬁdently prove a power–law distribution. The fact that both data sets show a power law indicates that the observed power–law distribution extends over a wider range than the individual

STEPWISE ACCUMULATION OF MELT OR CONNECTED NETWORKS?

analyses. It is, however, unclear whether the veins and dykes in the two data sets are of the same generation. The observed power–law width distributions are consistent with the model of Bons et al. (2004), but should also arise from the fractal-tree or fractal Menger–Sponge models of Tanner (1999), Weinberg (1999) and Maaløe (2003). A power–law distribution should, however, not be equated with a fractal distribution. Whereas the data do show power–law width distributions, the analyses of Marchildon & Brown (2003) show that the melt veins do not have a fractal geometry.

4. Discussion and conclusions Several arguments have been presented in the present paper that speak against the RFR for melt accumulation. The main argument is that a large-scale percolating network is unlikely to form in partially molten rocks, because (1) melt will start to ﬂow in smaller connected clusters, controlled by ductile ﬂow of the matrix, and (2) melt-ﬁlled fractures have a limited stability and can propagate driven by the melt buoyancy or applied stress gradients. Both processes have the eﬀect of destroying connectivity before a percolating network forms. Both processes also lead to accumulation of melt in ever-larger volumes. The ﬁeld evidence that has been used to support melt-ﬁlled networks has also been considered, focusing on the migmatite outcrops at Port Navalo, subject of studies by Brown and co-workers (Marchildon & Brown 2003; Brown 2004, 2005a). It has been shown that the apparent networks actually consist of diﬀerent generations of melt-ﬁlled veins and dykes. Crosscutting relationships indicate that previous generations of melt structures had already solidiﬁed when intruded by younger generations. The interpretation of discordant leucosomes as dilatant structures that form part of the network must also be reconsidered, as these are more probably contraction structures formed by the escape of melt. Both the theoretical arguments and the ﬁeld evidence speak against percolating melt networks in which melt ﬂows from little rivulets (leucosomes) to large rivers, the dykes that feed plutons. Instead, the present authors argue for a stepwise accumulation process, whereby batches of melt interact and accumulate locally (e.g. Maaløe 1987; Bons & van Milligen 2001; Bons et al. 2001a, b, 2004). Whereas the whole process of melt accumulation, and ﬁnally extraction, may span the entire thermal phase, individual accumulation events take place on a much shorter time scale. This model is consistent with the observed collapse structures found in the ﬁeld, which represent sites where batches of melt once resided, but which have now moved elsewhere. The power–law distributions of leucosome widths, observed in several migmatite terrains, is also consistent with a stepwise accumulation of melt batches (Bons et al. 2004).

5. Acknowledgements This project was partly funded by a German Research Foundation grant to Bons and Becker (DFG-project BO1776/4). Urtson acknowledges travel support by the German DAAD. We are grateful to Mike Brown for introducing us to the migmatite outcrops at Port Navalo. The manuscript beneﬁted from constructive reviews by Mike Brown and Nick Petford, and editorial handling by John Clemens.

6. References Allibone, A. H. & Norris, R. J. 1992. Segregation of leucogranite microplutons during syn-anatectic deformation: an example from

113

the Taylor valley, Antarctica. Journal of Metamorphic Geology 10, 589–600. Bonnet, E., Bour, O., Odling, N. E., Davy, P., Main, I., Cowie, P. & Berkowitz, B. 2001. Scaling of fracture systems in geological media. Reviews of Geophysics 39, 347–83. Bons, P. D. 1999. Apparent extensional structures due to volume loss. Proceedings of the Estonian Academy of Sciences, Geology 48, 3–14. Bons, P. D., Dougherty-Page, J. & Elburg, M. A. 2001a. Analogue modelling of segregation and ascent of magma (on-line). Journal of the Virtual Explorer 3: Animations in Geology – visualising the Earth. Bons, P. D., Dougherty-Page, J. & Elburg, M. A. 2001b. Stepwise accumulation and ascent of magmas. Journal of Metamorphic Geology 19, 627–33. Bons, P. D., Arnold, J., Elburg, M. A., Kalda, J., Soesoo, A. & van Milligen, B. P. 2004. Melt extraction and accumulation from partially molten rocks. Lithos 78, 25–42. Bons, P. D. & van Milligen, B. P. 2001. A new experiment to model self-organized critical transport and accumulation of melt and hydrocarbons from their source rocks. Geology 29, 919–22. Brown, M. 1983. The petrogenesis of some migmatites from the Presqu’ile de Rhuys. In Atherton, M. P. & Gribble, C. (eds) Migmatites, Melting and Metamorphism, 174–200. Nantwich: Shiva Publications. Brown, M. A. 1994. The generation, segragation, ascent and emplacement of granite magma: the migmatite- to crustally-derived granite connection in thickened orogens. Earth-Science Reviews 36, 83–130. Brown, M. A. 2004. The mechanism of melt extraction from lower continental crust of orogens. Transactions of the Royal Society of Edinburgh: Earth Sciences 95, 35–48. Brown, M. A. 2005a. Synergistic eﬀects of melting and deformation: an example from the Variscan belt, western France. In Gapais, D., Brun, J. P. & Cobbold, P. R. (eds) Deformation Mechanisms, Rheology and Tectonics: from Minerals to the Lithosphere. Geological Society, London, Special Publication 243, 205–26. Bath, UK: The Geological Society Publishing House. Brown, M. A. 2005b. ‘Comment’ in Brown, M. A., Pitcher, W. S., Vernon, R. H. & Clemens, J. D. Invited comments on Clemens’s ‘Granites and granitic magmas’. Proceedings of the Geologists’ Association 116, 17–32. Brown, M., Averkin, Y. A., McLellan, E. L. & Sawyer, E. W. 1995. Melt segregation in migmatites. Journal of Geophysical Research 100, 15655–79. Brown, M. A., Brown, M., Carlson, W. D. & Denison, C. 1999. Topology of syntectonic melt-ﬂow networks in the deep crust: Inferences from three-dimensional images of leucosome geometry in migmatites. American Mineralogist 84, 1793–818. Brown, M. & Dallmeyer, D. 1996. Rapid Variscan exhumation and the role of magma in core complex formation: southern Brittany metamorphic belt, France. Journal of Metamorphic Geology 14, 361–79. Brown, M. & Solar, G. S. 1998. Shear-zone systems and melts: feedback and self-organization in orogenic belts. Journal of Structural Geology 20, 211–27. Clemens, J. D. 1998. Observations on the origins and ascent mechanisms of granitic magmas. Journal of the Geological Society, London 155, 843–51. Clemens, J. D. & Mawer, C. K. 1992. Granitic magma transport by fracture propagation. Tectonophysics 204, 339–60. Druguet, E. & Carreras, J. 2006. Analogue modelling of syntectonic leucosomes in migmatitic schists. Journal of Structural Geology 26, 1734–47. Elburg, M. A. 1996. Evidence of isotopic equilibration between microgranitoid enclaves and host granodiorite, Warburton Granodiorite, Lachlan Fold Belt, Australia. Lithos 38, 1–22. Ellis, D. J. & Obata, M. 1992. Migmatite and melt segregation at Cooma, New South Wales. Transactions of the Royal Society of Edinburgh: Earth Sciences 83, 95–106. Emerman, S. H., Turcotte, D. L. & Spence, D. A. 1986. Transport of magma and hydrothermal solutions by laminar and turbulent ﬂuid fracture. Physics of Earth and Planetary Interiors 41, 249–59. Emerman, S. H. & Marrett, R. 1990. Why dykes? Geology 18, 231–3. Exner, U., Mancktelow, N. S. & Grasemann, B. 2004. Progressive development of s-type ﬂanking folds in simple shear. Journal of Structural Geology 26, 2191–201. Gapais, D., Lagarde, J. L., Lecorre, C., Audren, C., Jegouzo, P., Sainz, A. C. & Vandendriessche, J. 1993. The Quiberon Shear Zone – Evidence for Carboniferous Extension in the Variscan Belt

114

PAUL D. BONS ET AL.

of South Brittany (France). Comptes Rendus de l’Academie des Sciences Serie II 316, 1123–9. Gomez-Rivas, E., Bons, P. D., Griera, A., Carreras, J., Druguet, E. & Evans, L. 2007. Strain and vorticity analysis using small-scale faults and associated drag folds. Journal of Structural Geology 29, 1882–99. Grasemann, B. & Stu¨we, K. 2001. The development of ﬂanking folds during simple shear and their use as kinematic indicators. Journal of Structural Geology 23, 715–24. Harris, N., Vance, D. & Ayres, M. 2000. From sediment to granite: timescales of anatexis in the upper crust. Chemical Geology 162, 155–67. Hart, S. R. 1993. Equilibrium during mantle melting: a fractal tree model. Proceedings National Academy of Sciences USA 90, 11914–18. Hollister, L. S. & Crawford, M. L. 1986. Melt-enhanced deformation: A major tectonic process. Geology 14, 558–61. Holtzman, B. K., Groebner, N. J., Zimmerman, M. E., Ginsberg, S. B. & Kohlstedt, D. L. 2003a. Stress-driven melt segregation in partially molten rocks. Geochemistry, Geophysics, Geosystems 4, 8607. Holtzman, B. K., Kohlstedt, D. L., Zimmerman, M. E., Heidelbach, F., Hiraga, T. & Hustoft, J. 2003b. Melt segregation and strain partitioning: implications for seismic anisotropy and mantle ﬂow. Science 103, 1227–30. Jackson, M. D., Cheadle, M. J. & Atherton, M. P. 2003. Quantitative modeling of granitic melt generation and segregation in the continental crust. Journal of Geophysical Research 108(B7), 2332– 53. Johannes, W., Ehlers, C., Kriegsman, L. M. & Mengel, K. 2003. The link between migmatites and S-type granites in the Turku area, southern Finland. Lithos 68, 69–90. Johnson, T. M. & Brown, M. 2004. Quantitative constraints on metamorphism in the Variscides of Southern Brittany – a complementary pseudosection approach. Journal of Petrology 45, 1237–59. Kisters, A. F. M., Gibson, R. L., Charlesworth, E. G. & Anhaeusser, C. R. 1998. The role of strain localization in the segregation and ascent of anatectic melts, Namaqualand, South Africa. Journal of Structural Geology 20, 229–42. Koukouvelas, I. K. & Kokkolas, S. 2003. Emplacement of the Miocene west Naxos pluton (Aegean Sea, Greece): a structural study. Geological Magazine 140, 45–61. Kriegsman, L. M. 2001. Quantitative ﬁeld methods for estimating melt production and melt loss. Physics and Chemistry of the Earth 26, 247–53. Kruhl, J. H. 1994. The formation of extensional veins: an application of the Cantor-dust model. In Kruhl, J. H. (ed.) Fractals and dynamic systems in geoscience, 95–104. Berlin: Springer Verlag. Maaløe, S. 1987. The generation and shape of feeder dykes from mantle sources. Contributions to Mineralogy and Petrology 96, 47–55. Maaløe, S. 2003. Melt dynamics of a partially molten mantle with randomly oriented veins. Journal of Petrology 44, 1193–210. Marchildon, N. & Brown, M. 2003. Spatial distribution of meltbearing structures in anatectic rocks from Southern Brittany, France: implications for melt transfer at grain- to orogen-scale. Tectonophysics 364, 215–35. Matte, P. 2001. The Variscan collage and orogeny (480–290 Ma) and the tectonic deﬁnition of the Armorica microplate: a review. Terra Nova 13, 122–8. Me´riaux, C., Lister, J. R., Lyakhovsky, V. & Agnon, A. 1999. Dyke propagation with distributed damage of the host rock. Earth and Planetary Science Letters 165, 177–85. Me´riaux, C. & Jaupart, C. 1998. Dike propagation through an elastic plate. Journal of Geophysical Research 103, 18295–314. Moyen, J.-F., Ne´de´lec, A., Martin, H. & Jayananda, M. 2003. Syntectonic granite emplacement at diﬀerent structural levels: the Closepet granite, South India. Journal of Structural Geology 25, 611–31. Nicolas, A. 1986. A melt extraction model based on structural studies in mantle peridotites. Journal of Petrology 27, 999–1022. Nicolas, A. & Jackson, M. 1982. High temperature dikes in peridotites: origin by hydraulic fracturing. Journal of Petrology 23, 568–82. Oliver, N. H. S. & Barr, T. D. 1997. The geometry and evolution of magma pathways through migmatites of the Halls Creek Orogen, Western Australia. Mineralogical Magazine 14, 3–14. Olson, S. N., Marsh, B. D. & Baumgartner, L. P. 2004. Modelling mid-crustal migmatite terrains as feeder zones for granite plutons: the competing dynamics of melt transfer by bulk versus porous

ﬂow. Transactions of the Royal Society of Edinburgh: Earth Sciences 95, 49–58. Paterson, S. R. & Fowler, T. K. 1993. Re-examining pluton emplacement processes. Journal of Structural Geology 15, 191–206. Paterson, S. R. & Vernon, R. H. 1995. Bursting the bubble of ballooning plutons: a return to nested diapirs emplaced by multiple processes. Geological Society of America Bulletin 107, 1356–80. Petford, N., Cruden, A. R., McCaﬀrey, K. J. W. & Vigneresse, J.-L. 2000. Granite magma formation, transport and emplacement in the Earth’s crust. Nature 408, 669–73. Petford, N. & Koenders, M. A. 1998. Self-organisation and fracture connectivity in rapidly heated continental crust. Journal of Structural Geology 20, 1425–34. Rosenberg, C. L. & Handy, M. R. 2001. Mechanisms and orientation of melt segregation paths during pure shearing of a partially molten rock analog (norcamphor–benzamide). Journal of Structural Geology 23, 1917–32. Rubin, A. M. 1995a. Propagation of magma-ﬁlled cracks. Annual Review of Earth and Planetary Science 23, 287–336. Rubin, A. M. 1995b. Why geologists should avoid using ‘fracture toughness’ (at least for dykes). In Baer, G. & Heimann, A. A. (eds) Physics and Chemistry of Dykes, 53–63. Rotterdam: Balkema. Rutter, E. H. & Neumann, D. 1995. Experimental deformation of partially molten Westerly granite under ﬂuid-absent conditions with implications for the extraction of granitic magmas. Journal of Geophysical Research 100, 15697–715. Sawyer, E. W. 1987. The role of partial melting and fractional crystallization in determining discordant migmatite leucosome compositions. Journal of Petrology 28, 445–73. Sawyer, E. W. 1996. Melt segregation and magma ﬂow in migmatites: implications for the generation of granite magmas. Transactions of the Royal Society Edinburgh: Earth Sciences 87, 85–94. Secor, D. T. & Pollard, D. D. 1975. On the stability of open hydrofractures in the Earth’s crust. Geophysical Research Letters 2, 510–13. Sleep, N. H. 1988. Tapping of melt by veins and dikes. Journal of Geophysical Research 93, 10255–72. Soesoo, A., Kalda, J., Bons, P. D., Urtson, K. & Kalm, V. 2004. Fractality in geology: a possible use of fractals in the studies of partial melting processes. Proceedings of the Estonian Academy of Sciences, Geology 53, 13–27. Soesoo, A. & Nicholls, I. A. 1999. Maﬁc rocks spatially associated with Devonian felsic intrusions of the Lachlan Fold Belt: a possible mantle contribution to crustal evolution processes. Australian Journal of Earth Sciences 46, 725–34. Solar, G. S. & Brown, M. 2001. Petrogenesis of migmatites in Maine, USA: possible source of peraluminous leucogranite in plutons? Journal of Petrology 42, 789–823. Spence, D. A., Sharp, P. W. & Turcotte, D. L. 1987. Buoyancy-driven crack propagation: a mechanism for magma migration. Journal of Fluid Mechanics 174, 135–53. Takada, A. 1990. Experimental study on propagation of liquid-ﬁlled crack in gelatin: shape and velocity in hydrostatic stress condition. Journal of Geophysical Research 95, 8471–81. Tanner, D. C. 1999. The scale-invariant nature of migmatite from the Oberpfalz, NE Bavaria and its signiﬁcance for melt transport. Tectonics 302, 297–305. Urtson, K. & Soesoo, A. 2007. An analogue model of melt segregation and accumulation processes in the Earth’s crust. Estonian Journal of Earth Sciences 56, 3–10. Vanderhaeghe, O. 2001. Melt segregation, pervasive melt migration and magma mobility in the continental crust: the structural record from pores to orogens. Physics and Chemistry of the Earth (A) 26, 213–23. Vigneresse, J. L. 2004. A new paradigm for granite generation. Transactions of the Royal Society of Edinburgh: Earth Sciences 95, 11–22. Vigneresse, J. L. 2007. The role of discontinuous magma inputs in felsic magma and ore generation. Ore Geology Reviews 30, 181–216. Vigneresse, J. L., Barbey, P. & Cuney, M. 1996. Rheological transitions during partial melting and crystallization with application to felsic magma segregation and transfer. Journal of Petrology 37, 1579–600. Vigneresse, J. L. & Burg, J. P. 2000. Continuous vs. discontinuous melt segregation in migmatites: insight from a cellular automaton model. Terra Nova 12, 188–92.

STEPWISE ACCUMULATION OF MELT OR CONNECTED NETWORKS? Walte, N. P., Bons, P. D. & Passchier, C. W. 2005. Deformation of melt-bearing systems – insight from in situ grain-scale analogue experiments. Journal of Structural Geology 27, 1666–79. Weertman, J. 1971. Theory of water-ﬁlled crevasses in glaciers applied to vertical magma transport beneath ocean ridges. Journal of Geophysical Research 76, 1171–83. Weinberg, R. F. 1999. Mesoscale pervasive felsic magma migration: alternatives to dyking. Lithos 46, 393–410. Weinberg, R. F. & Podladchikov, Y. Y. 1994. Diapiric ascent of magmas through power law crust and mantle. Journal of Geophysical Research 99, 9543–59.

115

Weinberg, R. F. & Searle, M. P. 1998. The Pangong Injection Complex, Indian Karakoram: a case of pervasive granite ﬂow through hot viscous crust. Journal of the Geological Society, London 155, 883–91. Zorpi, M. J., Coulon, C., Orsini, J. B. & Cocirta, C. 1989. Magma mingling, zoning and emplacement in calc-alkaline granitoid plutons. Tectonophysics 157, 315–29.

MS received 18 October 2007. Accepted for publication 1 April 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 117–132, 2010 (for 2009)

Emplacement and assembly of shallow intrusions from multiple magma pulses, Henry Mountains, Utah Eric Horsman1*, Sven Morgan2, Michel de Saint-Blanquat3, Guillaume Habert3, Andrew Nugent2, Robert A. Hunter1 and Basil Tikoﬀ1 1

University of Wisconsin–Madison, Madison, Wisconsin, USA

*Current address: East Carolina University, Greenville, North Carolina, USA Email: [email protected] 2

Central Michigan University, Mt Pleasant, Michigan, USA

3

CRNS-LMTG, Observatoire Midi-Pyre´ne´es, Universite´ Paul-Sabatier, Toulouse, France

ABSTRACT: This paper describes three mid-Tertiary intrusions from the Henry Mountains (Utah, USA) that were assembled from amalgamation of multiple horizontal sheet-like magma pulses in the absence of regional deformation. The three-dimensional intrusion geometries are exceptionally well preserved and include: (1) a highly lobate sill; (2) a laccolith; and (3) a bysmalith (a cylindrical, fault-bounded, piston-like laccolith). Individual intrusive sheets are recognised on the margins of the bodies by stacked lobate contacts, and within the intrusions by both intercalated sedimentary wallrock and formation of solid-state fabrics. Finally, conduits feeding these intrusions were mostly sub-horizontal and pipe-like, as determined by both direct observation and modelling of geophysical data. The intrusion geometries, in aggregate, are interpreted to reﬂect the time evolution of an idealised upper crustal pluton. These intrusions initiate as sills, evolve into laccoliths, and eventually become piston-like bysmaliths. The emplacement of multiple magma sheets was rapid and pulsed; the largest intrusion was assembled in less than 100 years. The magmatic fabrics are interpreted as recording the internal ﬂow of the sheets preserved by fast cooling rates in the upper crust. Because there are multiple magma sheets, fabrics may vary vertically as diﬀerent sheets are traversed. These bodies provide unambiguous evidence that some intrusions are emplaced in multiple pulses, and that igneous assembly can be highly heterogeneous in both space and time. The features diagnostic of pulsed assembly observed in these small intrusions can be easily destroyed in larger plutons, particularly in tectonically active regions. KEY WORDS:

fabric analysis, laccolith, magma ﬂow, pluton emplacement, sill

A subtle but fundamental shift is occurring in our perception of how igneous bodies intrude into the crust. Speciﬁcally, it involves the concept of assembly, which recognises that all plutons may not have existed as a single large magma bodies (e.g. Coleman et al. 2004; Glazner et al. 2004). Rather, plutons may grow through amalgamation of sequentially intruded sheets (e.g. Mahan et al. 2003; Michel et al. 2008) or relatively small magma pulses with a variety of geometries (e.g. Matzal et al. 2006). A brief historical review provides the context for this issue. During the past two decades, much work has concentrated on a four-part sequence of magma generation, segregation, ascent, and emplacement (Petford et al. 2000). This sequence developed partially in response to the increasing lack of evidence for the upward movement of magma through the crust as diapirs. Diapiric transport of magma inherently combines both the ascent (vertical movement) and emplacement (transition to sub-horizontal movement) of magma bodies. The concept of diapirism is no longer generally regarded as a major ascent process in the upper crust, but the concept of intrusion of magma as a single large pulse lingers. While large magma bodies must exist to result in large-volume ignimbrite

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016089

ﬂows (e.g. Lipman 1984, 2007), there is no compelling reason to make the a priori assumption that all plutons were single magma bodies. For the purposes of this article, emplacement is described as the displacement of the surrounding rocks that allows a pluton to attain its three-dimensional geometry (Fig. 1). For this reason, emplacement mechanisms typically describe spacemaking mechanisms (e.g. roof lifting, ﬂoor depression, stoping), and it is generally recognised that multiple emplacement mechanisms facilitate the emplacement of any pluton (Hutton 1988, 1997; Paterson & Fowler 1993). Assembly is deﬁned as the process of pluton construction through magmatic addition. Assembly can occur in a single magmatic pulse or as an amalgamation of sequentially emplaced magma pulses. Within the context of these deﬁnitions, emplacement and assembly are diﬀerent concepts. A pluton emplaced by roof lifting could have been assembled from a single pulse of magma or a series of pulses (Fig. 1). A major problem remains: the assembly of plutonic bodies is often cryptic. Glazner et al. (2004) argue that the range of ages in the Tuolumne Intrusive Suite in the Sierra Nevada batholith, California, is larger than would be expected from a

118

ERIC HORSMAN ET AL.

Figure 1 Diagram illustrating the distinction between emplacement and assembly, as used in this paper. For a given observed emplacement geometry, multiple assembly histories are possible.

single, cooling magma body. Consequently, periodic inﬂuxes of magma are required to explain the observed trends. However, direct observation of multiple magma pulses is often diﬃcult. For example, recognition of multiple pulses within plutons is only straightforward when a major compositional diﬀerence exists (e.g. Wiebe & Collins 1998; Matzal et al. 2006). Similarly, primary fabrics within plutons can result from both emplacement-related and assembly-related processes, and separation of these eﬀects is often diﬃcult. Lastly, regional deformation can dramatically inﬂuence both pluton geometry and fabric patterns, making inferences of assembly history equivocal (Paterson et al. 1998). Results are presented of a detailed study of intrusions in the Henry Mountains of southern Utah, a location ideally suited for separating emplacement from assembly of igneous bodies. First, the primary emplacement mechanism for these small intrusions is roof lifting, as ﬁrst proposed by Gilbert (1877) and supported by subsequent workers (e.g. Hunt 1953; Johnson & Pollard 1973; Pollard & Johnson 1973; Jackson & Pollard 1988). Secondly, the intrusions exist at a shallow crustal level where cooling was suﬃciently rapid that magmatic fabrics and evidence of multiple magma pulses are preserved. Thirdly, the intrusions are exceptionally well exposed, as the surrounding sedimentary rocks are distinctly more susceptible to erosion than the igneous bodies. Finally, emplacement of the intrusions occurred on the Colorado Plateau during a time of tectonic quiescence, and therefore the fabrics are not aﬀected by regional deformation. The results demonstrate that assembly occurred by intrusion of a series of sub-horizontal igneous sheets that contain complex internal fabrics. These sheets are locally fed by

sub-horizontal tube-shaped conduits, which also exhibit evidence for multiple magmatic pulses. Evidence for magmatic pulses becomes increasingly cryptic as the size of the intrusion increases. Using this evidence, constraints are summarised from numerical modelling on the time scale of the magma intrusion. Finally, the paper discusses how assembly inﬂuences emplacement models and provides insights into how intrusions grow in the upper crust.

1. The Henry Mountains The mid-Tertiary igneous bodies of the Henry Mountains intrude the ﬂat-lying stratigraphy of the Colorado Plateau (Fig. 2). Displacement of the wallrock therefore directly records intrusion geometry. The intrusions post-date the minor Laramide-age deformation that aﬀected this region. Therefore, fabrics within the intrusion reﬂect emplacement processes and lack a tectonic overprint. The magmas have a geochemical signature typical of volcanic arcs above a subduction zone (Nelson et al. 1992; Nelson 1997; Saint Blanquat et al. 2006; Bankuti 2007) and are part of a diﬀuse pattern of simultaneous magmatism throughout the region, which is interpreted to reﬂect arc-like magmatism above a shallowly dipping slab (Nelson et al. 1992) ﬁltered through the thick crust of the Colorado Plateau (Thompson & Zoback 1979). Combined with the 3–4 km original depth of emplacement (Jackson & Pollard 1988), these features make the Henry Mountains an ideal location to study shallow igneous emplacement processes of arc magmas in a relatively simple system. Complications associated with regional tectonism, found in essentially all arcs, are absent.

EMPLACEMENT AND ASSEMBLY OF SHALLOW INTRUSIONS

119

Figure 2 (a) Simpliﬁed geologic map of the Henry Mountains region. Box shows the location of (b). Location of cross section shown in (c) is also indicated. Inset shows the location of (a) on the Colorado Plateau in the western United States cordillera. (b) Shaded relief map of the eastern portion of Mt Hillers. Outlines of the three intrusions and the conduit discussed in this paper are shown in white. (c) Schematic cross section through Mt Hillers, oriented NE–SW.

Five intrusive centres comprise the Henry Mountains (Fig. 2). Each intrusive centre is a large and complex laccolithic body (Jackson & Pollard 1988, 1990), made of dozens of interconnected component intrusions with a wide range of geometries. All three of the intrusions studied on the eastern margin of the Mt Hillers intrusive centre (Fig. 2b) are in a unique state of erosion. Numerous upper, lower and lateral contacts with the surrounding sedimentary rock are preserved. The level of detail of intrusion geometry preserved helps to illuminate the complex emplacement and assembly processes of the igneous bodies.

1.1. Composition, fabric, and wallrock deformation The intrusions are composed of plagioclase-hornblende porphyry whose bulk chemistry is consistently similar throughout the range (Nelson et al. 1992; Bankuti 2007). Primary phenocryst populations include 30–35% by volume 0·5–1·5 cm-diameter euhedral plagioclase laths, and 5–15% by volume 0·1–

0·5 cm-long hornblende needles. Other phenocrysts include <2% euhedral to subhedral oxides (principally magnetite) and <1% euhedral apatite and sphene. The matrix commonly makes up 50% or more of the rock and is composed primarily of very ﬁne-grained plagioclase, hornblende, and oxides. Field fabric is variably well developed in the intrusions (Morgan et al. 2005), as described in more detail for each body below. In general, fabrics are strongest near intrusion margins and decrease in intensity away from the margins. A carapace of solid-state fabric is locally preserved in the outermost few centimetres of each intrusion. Where the carapace is well preserved, it grades smoothly into magmatic fabric away from the wallrock contact. Asymmetry of the solid state and magmatic fabric relative to the contact provides a useful shear sense indicator (e.g. Horsman et al. 2005). Further from contacts, fabric is less well developed. To characterise these weak fabrics, the anisotropy of magnetic

120

ERIC HORSMAN ET AL.

Figure 3 (a) Geologic map of the Maiden Creek sill. Lineations shown with black arrows include ﬁeld measurements and magnetic fabric results, with arrow length inversely proportional to plunge. Locations of cross-sections shown in (b) are indicated. Large grey arrows indicate inferred primary magma ﬂow directions based on intrusion geometry and fabric patterns. UTM zone 11. (b) Cross-sections through the intrusion. Note that the scale of cross-section E–E# is diﬀerent from that of the others.

susceptibility (AMS) was measured at numerous stations in each intrusion. Where detailed comparisons have been made, AMS orientations agree well with both ﬁeld measurements and three-dimensional laboratory fabric analysis results (Horsman et al. 2005). The AMS signal in the porphyry is carried principally by multi-domain and pseudo-single-domain magnetite (Habert & de Saint Blanquat 2004; Horsman et al. 2005; Morgan et al. 2008), allowing for normal interpretation of results; the maximum eigenvector of the magnetic susceptibility tensor is the magnetic lineation and the minimum eigenvector is the pole to magnetic foliation. Both ﬁeld-based measurements and magnetic fabric data provide constraints on magma ﬂow patterns during emplacement. The intrusions are small and occur in an upper crustal setting, suggesting that they cooled quickly and do not record any structural overprinting. In fact, three-dimensional intrusion geometries are so well constrained that the ﬁrst-order magma ﬂow patterns can be inferred directly from the geometry and then tested against the observed fabrics. Aside from fabric analysis, wallrock deformation provides information on the emplacement and assembly history of each intrusion. Analysis of fault and fracture orientations and oﬀset sometimes allows for inference of detailed emplacement history (e.g. Pollard & Johnson 1973; Morgan et al. 2008). Interpretation of wallrock deformation and igneous fabrics together provides a more complete picture of emplacement and assembly history. Each data set records distinct processes and these data are integrated in the present paper to provide a more complete analysis of emplacement and assembly.

2. Maiden Creek sill 2.1. Three-dimensional geometry The Maiden Creek sill (Fig. 3) represents the initial stage of igneous emplacement with a relatively small volume of magma (<0·03 km3). A more thorough description of the intrusion in provided by Horsman et al. (2005). Numerous upper, lower and lateral contacts with wallrock sandstone are preserved. These contacts tightly constrain a complex three-dimensional geometry. In map view, the sill has an elliptical main body from which protrude several ﬁngerlike, 100 m-scale lobes. Similar lobate sill geometries have been observed in detailed three-dimensional seismic reﬂection studies in the North Sea (Thomson & Hutton 2004; Hansen & Cartwright 2006). In detail, the cross sectional geometry of the intrusion is similarly complex. Whilst the main body has a simple sill-like cross sectional geometry with concordant upper and lower contacts, cross sections along the length of the two bestexposed ﬁnger-like lobes (Fig. 3b) show that the base of the intrusion cuts up through the stratigraphic section. Along the easternmost ﬁnger this process was gradual (cross section A–A# on Fig. 3b), while along the SE ﬁnger the process was more step-like (cross section D–D# on Fig. 3b).

2.2. Internal fabric and wallrock deformation Solid-state fabric exists in the outermost few centimetres of the intrusion near every exposed contact with wallrock, and this

EMPLACEMENT AND ASSEMBLY OF SHALLOW INTRUSIONS

carapace is preserved in some locations where the sedimentary rock has eroded, suggesting that the erosion is relatively recent. This solid-state fabric grades smoothly into a similarly oriented magmatic fabric over a few centimetres. Further from the contact magmatic fabric becomes weaker. Foliation is consistently sub-horizontal and lineation has a fanning radial map pattern in the elliptical main body (Fig. 3). In each ﬁnger-like lobe, lineation has a more restricted fanning map pattern generally aligned with the long axis of the lobe. The tightly constrained intrusion geometry requires that magma ﬂowed from the main body into the lobes (Fig. 3).

2.3. Evidence for multiple magma pulses Several lines of evidence demonstrate that the sill is composed of two stacked horizontal sheets of magma that intruded sequentially. Several cross-sections through the lateral margin of the intrusion are preserved and everywhere show two stacked bulbous terminations (ﬁgs 4a and 4b in Horsman et al. 2005). Additionally, the sub-horizontal boundary between the two sheets is marked locally by either metre-scale lenses of intercalated sandstone or centimetre-scale zones of solid-state fabric. The combination of strong fabric at the intrusion margins and weak fabric in the interior suggests that a signiﬁcant portion of the emplacement-related strain in the intrusion was accommodated in the outer few centimetres of each sheet. This fabric also suggests that emplacement of each sheet occurred rapidly. The two sheets were probably emplaced in rapid succession, because the sheet-sheet contact is commonly diﬃcult to recognise, except where marked by intercalated wallrock. The well-preserved lateral margins of each sheet are bulbous in cross-section, rather than pointed as predicted by many mechanical models of sheet intrusions (e.g. Pollard 1973).

2.4. Summary The Maiden Creek intrusion is a sill with a complex geometry. It is lobate in map view and the bottom contact cuts upward in cross-section. Both solid-state and magmatic fabrics closely preserve magma ﬂow patterns. These fabric patterns correspond well with the complex intrusion geometry and demonstrate that magma ﬂowed from the main body out into the ﬁnger-like lobes to produce fanning lineation and arcuate foliation patterns that reﬂect progressive growth of the intrusion. The sill is composed of two stacked sheets that were emplaced sequentially as separate magma pulses.

3. Trachyte Mesa laccolith 3.1. Three-dimensional geometry The Trachyte Mesa laccolith (Fig. 4) has at least twice the volume of magma (w0·05 km3) as the Maiden Creek sill. The intrusion is elongate along a line pointing radially away from the main Mt Hillers intrusive centre. Morgan et al. (2008) provide a more detailed description of this intrusion. Numerous upper contacts of this intrusion with ﬂat-lying wallrock are well preserved, and lower contacts are locally preserved. These contacts indicate that the intrusion has a broadly sill-like cross-sectional geometry (Fig. 4b). A well-preserved lateral contact on the NW margin indicates that the intrusion grew in a laccolithic manner. In this location, wallrock layers are bent upward and thinned over the intrusion. Strain analyses of these sandstone layers demonstrate that deformation of the overlying rock was localised at the intrusion margins through vertical growth of the intrusion.

121

There is also evidence for multiple (>10) stacked igneous sheets at this margin, identiﬁed by bulbous terminations, solid-state fabrics and intercalated sandstone. Cross-cutting relationships between sheets demonstrate a complex stacking history, but are consistent with a general pattern of emplacement of younger sheets below older sheets. For example, some of these igneous sheets at the margin are rotated with the sedimentary layering, suggesting that they were intruded horizontally and rotated by subsequent emplacement of underlying sheets. However, late horizontal igneous sheets locally intrude the rotated sedimentary beds and earlier igneous sheets. Although the main body of the intrusion is essentially laccolithic, at several locations on the margins of the intrusion complex, ﬁnger-like lobes are preserved (Fig. 4). These 100-m-scale, highly elongate lobes commonly project out from the margin of the main Trachyte Mesa intrusion. These lobes are similar to the ﬁnger-like lobes observed within the Maiden Creek intrusion. However, the lobes make up only a small portion of the Trachyte Mesa intrusion volume instead of the majority they make up on the Maiden Creek intrusion.

3.2. Internal fabric and wallrock deformation Field and magnetic foliation data for the Trachyte Mesa intrusion are consistently sub-horizontal (Fig. 4). Lineation plunges are also consistently sub-horizontal, but trend values have a bilaterally symmetric geometry. Along a central axis, lineation is parallel to the length of the intrusion. Away from that axis, lineation trends perpendicular to the axis and points toward the intrusion margins. Flow patterns with this bilaterally symmetric geometry have traditionally been ascribed to ﬂow away from a central dike-like feeder below the intrusion (e.g. Ferre´ et al. 2002). However, data is presented (see section 5) indicating that the intrusion is fed by a sub-horizontal conduit from the direction of the main Mt Hillers intrusive centre. The average foliation on top of the intrusion also dips approximately 10( to the NE (Fig. 4a), parallel to the long axis of the intrusion, and is used as a bulk magma ﬂow indicator (Morgan et al. 2008). Therefore, it is suggested that this pattern represents magma ﬂow out along the length of the intrusion in a sub-horizontal conduit, which then feeds magma out to the lateral margins (see Morgan et al. 2008). A similar ﬂow pattern has been inferred for other sills (Thomson & Hutton 2004; Hansen & Cartwright 2006). Wallrock deformation is documented by Morgan et al. (2008). Little deformation is observed on the few outcrops of wallrock on the top of the intrusion, suggesting primarily vertical translation of these rocks. Deformation at the exposed lateral contact on the NW side of the intrusion is consistent with the sedimentary layering being translated upward and over the top of an upward- and outward-propagating hinge. Sedimentary layers immediately above the contact locally show evidence for layer-parallel shortening, which is inferred to result from coupling with the underlying sheets as they moved outward. Further from the contact, sedimentary layers show up to 20% layer-parallel thinning due to bulk roof lifting, accommodated by fracture-induced porosity collapse and faulting. These zones are decoupled from each other by a thin layer of intensely sheared sandstone.

3.3. Evidence for multiple magma pulses Although cross-cutting relationships clearly demonstrate the importance of pulsed assembly at the margins of the Trachyte Mesa intrusion, exposures of the interior have only equivocal evidence of sheets. In particular, the top of the intrusion has two distinct plateau-like levels, which are thought to

122

ERIC HORSMAN ET AL.

Figure 4 (a) Geologic map of the Trachyte Mesa laccolith. Lineations shown with black arrows are magnetic fabric results and all have sub-horizontal plunges. Equal-area lower hemisphere plots show orientations of magnetic lineation and poles to foliation. Note that foliation dips shallowly to the NE, parallel to the length of the intrusion. Grey arrows indicate inferred primary magma ﬂow directions based on intrusion geometry and fabric patterns. Numbered lines SW of the intrusion indicate locations of magnetic anomaly traverses described in Figure 7. UTM zone 11. (b) Cross-section through the intrusion. 2·5 times vertical exaggeration.

correspond to diﬀerent sheets. This inference is corroborated by slightly diﬀerent fabric patterns that occur at those diﬀerent levels. The lack of distinct sheet–sheet contacts suggests assembly was rapid enough for emplaced pulses to stay above the solidus before the next sheet intruded. These observations can be interpreted in a variety of ways. First, late-stage magmatic processes may have obscured

evidence of earlier pulsed assembly. Alternatively, sheets could form only near the margins of the intrusion while the interior was a single magma body. The latter hypothesis is not consistent with the emplacement, rotation, and pre-intrusion of sheets on the margins or the diﬀerent ﬂow ﬁelds on top of the intrusion. Consequently, it is inferred that late magmatic processes obscured evidence of earlier pulsed assembly.

EMPLACEMENT AND ASSEMBLY OF SHALLOW INTRUSIONS

123

Figure 5 (a) Geologic map of the Black Mesa intrusion. Lineations shown with black arrows are magnetic fabric results, with plunge indicated. Lineations shown with black lines are ﬁeld measurements of lineation trend. Grey arrows indicate primary magma ﬂow directions based on intrusion geometry and fabric patterns. UTM zone 11. (b) Cross-sections through the intrusion, with schematic trace of foliation indicated.

3.4. Summary The Trachyte Mesa laccolith is an elongate body in map-view. In cross-section it exhibits sub-horizontal, sill-like upper and lower contacts, along with laccolithic margins. Magma in the intrusion ﬂowed out along a central axial conduit that fed ﬂow to the lateral margins. Numerous magma pulses contributed to assembly of the intrusion. Evidence of these pulses as distinct sheets is clear at the lateral margins of the intrusion, but cryptic in the centre of the body.

4. Black Mesa bysmalith 4.1. Three-dimensional geometry The relatively voluminous (w0·4 km3) Black Mesa bysmalith (Fig. 5) is the largest of the studied intrusions. Habert & Saint Blanquat (2004) and Saint Blanquat et al. (2006) provide more detailed descriptions of this intrusion. As with the other two intrusions studied, numerous preserved top and lateral contacts allow the geometry of the intrusion, which is w1·7 km in

124

ERIC HORSMAN ET AL.

diameter and roughly circular in map view, to be clearly constrained. The eastern margin is a sub-vertical fault along which the sub-horizontal wallrock is displaced upward w200 m. This vertical oﬀset is determined from stratigraphic relationships between sub-horizontal shale and sandstone on the top of the intrusion and similar wallrock exposed at the base of the intrusion. This type of relation occurs on the margin of a bysmalith, which is deﬁned as an intrusion with a ﬂat ﬂoor and ﬂat roof that lifts the overburden in a piston-like manner along a curved, steeply dipping fault. In contrast, the western margin of the intrusion is a monocline. Taken together, the intrusion resembles a trapdoor, with a hinge along the western side and a discrete, faulted oﬀset along the eastern side. Consequently, the Black Mesa intrusion represents a transitional stage of development between a laccolith and a bysmalith.

4.2. Internal fabric and wallrock deformation There is little information available about the internal fabric of the Black Mesa intrusion, because the current exposures record almost exclusively the outermost margins of the body. Along the steep, faulted eastern margin of the intrusion, solid-state and magmatic lineations are both steeply plunging. Immediately adjacent to wallrock contacts, the fabric is solidstate and deﬁned by cataclastically deformed phenocrysts. Further from this contact, the fabric is magmatic and deﬁned by phenocryst shape-preferred orientation. The observed fabrics are consistent with syn-magmatic shearing on the sub-vertical bounding fault along the eastern margin of the intrusion. Where exposure exists, the top of the intrusion is also characterised by both solid-state and magmatic fabrics. Solidstate fabrics occur in the outer few centimetres and grade inward to magmatic fabrics, similar to the other intrusions. Foliation is sub-horizontal in both the solid-state and magmatic fabrics. Lineation orientation changes with distance from the top contact. The solid-state and magmatic lineations at the upper contact are oriented WNW, while the magmatic lineations are oriented NNE several centimetres below the contact. Saint Blanquat et al. (2006) interpret this fabric to indicate NNW stretching during divergent ENE-oriented magma ﬂow. They attribute the pattern to intrusion from below, perhaps by a dike-like feeder striking approximately NNE. The similarity of this fabric pattern to that observed in Trachyte Mesa intrusion suggests that a lateral pipe-like feeder at the base of the intrusion is also possible. However, no feeder conduit, is exposed. Outcrops of wallrock on the top of the intrusion are relatively undeformed. Minor fracturing exists as conjugate sets, with orientations generally consistent with regional trends away from the Henry Mountains. This observation is similar to overburden rocks atop both the Maiden Creek and Trachyte Mesa intrusions.

4.3. Evidence for multiple magma pulses Several observations provide indirect evidence of multiple pulses. In the upper portion of the intrusion, sub-horizontal textural layering and cataclastic zones can be traced for tens of metres. The remainder of the intrusion displays no internal contacts, although exposures of the interior are very rare. Additionally, petrologic and geochemical zonation exists along vertical proﬁles through the intrusion (Bankuti 2007). None of these observations provides unequivocal evidence of multiple magma pulses. However, together they strongly suggest that the intrusion was amalgamated from multiple magma pulses.

4.4. Summary The Black Mesa intrusion has an approximately cylindrical geometry. The western margin of the intrusion is monoclinal, while the eastern margin is a vertical fault with a displacement of w200 m. Magma in the intrusion ﬂowed out along a central axial conduit before ﬂowing laterally out to the margins. Multiple pulses of magma likely contributed to the assembly of the intrusion, but evidence for these pulses is more cryptic than in the two smaller intrusions studied.

5. Magma conduits Dikes are rare in the Henry Mountains. Instead, isolated sub-horizontal igneous bodies with approximately elliptical cross sections are observed, similar to the elongate lobes of the Maiden Creek body. Following Hunt (1953), it is suspected these were pipe-like feeders that locally provided magma for the separate intrusive bodies. Two examples are described below: (1) an exposed example of one of these feeders (section 5.1); and (2) geophysically imaged conduits that act as feeders to the Trachyte Mesa intrusion (section 5.2). In the absence of a purely descriptive geometric name deﬁned in the literature, these igneous bodies are referred to as conduits. These conduits presumably feed magma from the main Mt Hillers intrusive centre to the satellite bodies and represent the ﬁnal stage of a longer ascent process that principally involves vertical movement.

5.1. Black Canyon conduit Approximately 5 km SW of the Trachyte Mesa intrusion, a highly elongate igneous body is exposed for w150 m as a series of isolated outcrops in Black Canyon (Fig. 6). The composition and texture are similar to the larger igneous bodies in the area. This intrusion was originally reported by Hunt (1953), and is herein referred to as the Black Canyon conduit. For several of the individual outcrops, both the top and side contacts of the Black Canyon conduit are well exposed (Fig. 6). In general, the top contact is sub-horizontal, while the lateral contacts are sub-vertical. These observations constrain the width of the intrusion to less than w5–10 m along the length of the exposure. The absence of continuous outcrop indicates that the intrusion cannot be a continuous tabular body (e.g. a dike or sill). Rather, the intersection of the topography and the igneous body is most consistent with a pipe-like conduit, with the base of the intrusion oriented near the current exposure level of the body. Consequently, the Black Canyon conduit likely represents a pipe-like feeder transporting magma radially away from the main Mt Hillers intrusive centre. Hunt (1953) noted the Black Canyon conduit was located between the Trachyte Mesa intrusion and the Mt Hillers intrusive centre, with an orientation toward Trachyte Mesa. Hunt consequently suggested that the conduit might be an exposure of the feeder for the Trachyte Mesa intrusion. The present authors disagree with this interpretation for two principal reasons. First, the conduit is located higher in the stratigraphic section than the Trachyte Mesa intrusion. It is unlikely that a magma conduit would cut downsection, particularly given our observation of the upward-sloping bottom contact in the ﬁnger-like lobes of the Maiden Creek intrusion. Secondly, as described below (see section 5.2), geophysical studies suggest that multiple tube-like conduits feed the Trachyte Mesa intrusion below the current level of exposure. Fabric in the conduit is magmatic in the middle of the body and solid-state on the edges (Fig. 6). Magmatic fabrics farther from the wallrock contact are dominated by lineation, with

EMPLACEMENT AND ASSEMBLY OF SHALLOW INTRUSIONS

125

rates or suﬃcient cooling of the intrusion to cause fracturing. Kinematic indicators present within the breccia zones potentially indicate oblique slip, but this inferred shear sense varies from vein to vein. These structures are interpreted as recording the relative displacement between individual magma pulses. In summary, the Black Canyon conduit records dominantly linear fabrics in a tube-like body, suggesting outward magma ﬂow from the Mt. Hillers intrusive centre. The observed breccia veins, ﬁeld fabrics, and AMS results are consistent with the intrusion of multiple magma pulses through the conduit. The complicated lineation pattern likely records simultaneous magma ﬂow and intrusion growth within this small, geometrically complex body.

5.2. Trachyte Mesa feeder

Figure 6 (a) Simpliﬁed geological map of the Black Canyon conduit. Field-measured lineation shown at the station locations. AMS lineation and foliation results, plotted nearby in white boxes, are from the same stations. UTM zone 11. (b) Field photograph of several generations of breccia veins. Composition of clasts in the breccia include porphyry, sandstone, and older breccia.

local, poorly developed foliation. In general, solid-state and magmatic lineations near wallrock contacts trends perpendicular to the length of the intrusion, while magmatic lineation further from contacts trends roughly parallel to the length of the intrusion. The ﬁrst lineation, oriented NS–SE and subparallel to the long axis of the intrusion, is interpreted as recording ﬂow along the length of the conduit. Magmatic ﬂow presumably went from the SW to the NE, away from the Mt Hillers intrusive centre. The second lineation is interpreted as recording the radial expansion of the conduit during pulsed growth of the body. Six oriented samples were collected for AMS analysis. Samples were taken from rocks containing only magmatic fabrics and were processed similarly to those from the other intrusions (e.g. Horsman et al. 2005). Similar to the ﬁeld lineation, AMS results display an irregular lineation in the magnetic fabric that varies in both trend and plunge along the length of the intrusion (Fig. 6a), although the ﬁeld and magnetic lineation locally disagree. Magnetic foliation is generally ﬂat-lying, consistent with the one observed ﬁeld foliation. Magnetic fabrics are dominantly plane strain, which appears inconsistent with the strong ﬁeld-based linear fabric. Veins of ﬁne-grained, green breccia are locally preserved along the outcrop (Fig. 6b). These roughly planar veins generally strike sub-parallel to the trend of the intrusion and contain clasts of both porphyry and sandstone. Evidence of multiple cross-cutting diking events suggests multiple episodes of brecciation. The presence of these zones suggests multiple injections of magmas, with either very high strain

Magnetic anomaly data were collected along ﬁve traverses on the alluvial plateau SW of the Trachyte Mesa intrusion to provide constraints on the feeder system for the body. Data are presented from three of these traverses (1, 3, 5) in Figure 7 and locations are shown on Figure 4. Data from the two traverses not presented (2, 4) show patterns similar to the discussed traverses. The length and placement of the traverses were governed by the size of the accessible plateau area, eliminating any changes in the magnetic ﬁeld that might occur due to changes in elevation. The traverses were oriented NW–SE, perpendicular to an axis connecting Mt Hillers to the intrusion. The closest traverse is 110 m SW from the nearest exposure of the intrusion (Fig. 4). The ﬁrst and third traverses are 170 m long. The ﬁfth traverse, which is furthest from the intrusion, is 290 m long and is w320 m from the SW edge of the intrusion. Magnetic data were collected every three metres along each traverse using two magnetometers separated on a pole in the gradiometer mode. The data from each traverse were smoothed. Traverses one and three each reveal two to four anomalies, two of which can be traced between the traverses. Each of these two traverses has a relatively large anomaly near the NW end (Fig. 7). Traverse ﬁve, the most distant from the exposed intrusion, has numerous smaller anomalies. The magnitude of the anomalies varies considerably between the ﬁve traverses. Traverse one has anomalies that vary between positive and negative 10 /m, whereas traverse three has a positive anomaly close to 100 /m and traverse ﬁve has negative anomalies of similar magnitude. To test possible intrusion geometries, the observed anomaly proﬁles were forward modelled as dikes, sills, and ﬁnger-like bodies using GM-SYS software. Each 2·5-dimensional model simulates a block of country rock Entrada Sandstone that was intruded by various shapes of porphyry. The susceptibility and remanence of samples from the porphyry and from the country rock Entrada Sandstone were determined using rock magnetic facilities at Lake Superior State University and the University of Michigan. The sills and ﬁnger-like bodies were modelled to be at a depth of 12 m below the surface, based on the elevation of the observed base to the intrusion 300 m to the NE. Shapes of the modelled ﬁnger-like bodies were modiﬁed in order to have the greatest correspondence between the observed and calculated (model) magnetic intensities. Dikes were placed at various positions and with various dips for the best ﬁt. Dike widths and sill thicknesses were also modiﬁed for the best ﬁt. The models illustrate that along each traverse, a series of ﬁnger-like bodies more accurately matches the observed anomalies than one sill or a set of dikes. The models also show that a thick ﬁnger-like shape of porphyry, somewhat similar to the projected cross-sectional shape of the Black Canyon conduit, may exist on the NW end of the closest two traverses.

126

ERIC HORSMAN ET AL.

Figure 7 Magnetic anomaly data from traverses SW of the Trachyte Mesa laccolith. Forward models of the data test possible feeder geometries for each magnetic anomaly traverse. Tested feeder geometries include dikes, sills and ﬁnger-like bodies. Finger-like feeder bodies best match the observed anomaly data. Note that scales and vertical exaggeration diﬀer for each plot. Locations of the traverses are shown on Figure 4.

Correlation from traverse to traverse suggests that a network of ﬁnger-like lobes may anastomose at a particular bedding horizon and ﬁnally converge near the SW exposure of the Trachyte Mesa intrusion.

6. Discussion The geometries of the three intrusions studied lie along the hypothesised evolution of an upper crustal intrusion from sill to laccolith to bysmalith. Individually, each intrusion provides insight into diﬀerent aspects of shallow magma emplacement. Together, the intrusions provide a framework to consider several aspects of emplacement and assembly, including: (1) evolution of pluton geometry; (2) fabric patterns and magma ﬂow evolution; (3) times scales of emplacement and assembly; and (4) heterogeneity of magma pulses in space and time.

6.1. Evolution of pluton geometry The exceptional preservation of intrusion geometry in the Henry Mountains provides an informative view of the geometries of the three igneous bodies. Using observations of the intrusions as a guide, a hypothetical history of pluton assembly is presented during growth from sill to laccolith to bysmalith. Figure 8 shows idealised intrusion geometries for a sill, laccolith, and bysmalith with pulsed assembly histories. Figure 9 schematically shows both the primary emplacement mechanism (vertical displacement of the wallrock) and the preferred assembly histories for the three intrusions studied.

The Maiden Creek intrusion demonstrates the early stages of sill formation, which presumably involved a relatively small magma pulse spreading away from a pipe-like feeder in the form of ﬁnger-like lobes. In the immediate vicinity of the feeder, these lobes coalesced to form a sill-like body with a roughly elliptical shape in map view. A second magma pulse of approximately the same volume followed the same emplacement path, so that two stacked sheets comprise the intrusion. The Trachyte Mesa intrusion demonstrates how a pluton can evolve into a sill-like body and begin to inﬂate into a laccolith. Further magma pulses (or larger initial pulses) spread beyond the nascent sill observed at the Maiden Creek intrusion to produce a larger and more mature sill. This more developed intrusion maintains the tabular cross-sectional geometry, but exhibits a more simple map-view geometry, in which ﬁnger-like lobes constitute a small portion of the total volume. Late magma pulses followed emplacement paths similar to early pulses. Additionally, the pulses thickened the intrusion a few metres at a time. Eventually, earlier emplaced sheets were rotated upward and sometimes intruded by subsequent sheets to form a laccolithic margin. Assembly of the Trachyte Mesa intrusion ceased at this stage with characteristics of both sill and laccolith geometries. The Black Mesa intrusion illustrates further development of an igneous body with increased magma supply. As the number of emplaced sheets grew, the margins of the laccolith steepened. Eventually, a fault formed along the E margin of the

EMPLACEMENT AND ASSEMBLY OF SHALLOW INTRUSIONS

127

Figure 8 Schematic block diagrams of idealised intrusions, including magma sheets, inferred primary magma ﬂow directions and relationships with sedimentary wallrock. These diagrams show characteristic features of intrusions with sill, laccolith and bysmalith geometries assembled from multiple pulses. Although the sketches are somewhat similar to the observed intrusions in the Henry Mountains, they do not accurately reﬂect those observations.

laccolith and the overburden was lifted in a piston-like manner to produce a bysmalith. The circular map pattern minimises the area of the intrusion-bounding fault, maximising the eﬃciency of overburden uplift. Additionally, the diﬀerent geometries on the eastern and western margins of the Black Mesa intrusion provide clear evidence of the transitional nature of the change from laccolith to bysmalith. Emplacement of additional magma pulses would presumably have produced further roof uplift and eventually a fault around the entire pluton margin. Examples of well-developed bysmaliths are seen elsewhere in the Henry Mountains (e.g. Bull Mountain and Table Mountain on the Mt Ellen intrusive centre; Hunt 1953). The main distinctions between the diﬀerent types of intrusion appear to be (1) the amount of magma that is emplaced and (2) the manner in which the wallrock accommodates emplacement. The amount of magma appears to correspond to the speciﬁc number of magma pulses, recognisable as horizontal sheets at the periphery of the intrusion where cooling is rapid. The magma pulses must have similar volumes, within an order of magnitude, as they have similar thicknesses over approximately the same map area. The Maiden Creek intrusion consists of two magma pulses, the Trachyte Mesa intrusion consists of at least ten magma pulses, and thermal

modelling suggests the Black Mesa consists of at least ten pulses (Saint Blanquat et al. 2006). The Black Mesa intrusion also demonstrates the importance of the evolution of emplacement mechanisms as pluton volume grows. In fact, this intrusion preserves the transition of primary emplacement mechanism from wallrock rotation and distortion seen on the laccolithic western margin to wallrock uplift through faulting, as seen at the bysmalithic eastern margin. This transition is likely related to mechanical limits on the amount of strain the wallrock can accommodate, and perhaps to rates of deformation, with both high strain and high strain rates leading to localisation and faulting. During the progression from sill to laccolith to bysmalith, the volumetric importance of peripheral lobes decreases as total pluton volume increases. For example, a large portion of the intrusive volume of the Maiden Creek sill occurs as lobes radiating away from a relatively small central main body. The Trachyte Mesa laccolith, ﬁnger-like lobes exist only at the margin of the relatively large main body of the intrusion. No such lobes are exposed as part of the approximately cylindrical Black Mesa intrusion. This decrease in the volumetric importance of ﬁnger-like lobes may reﬂect increasing stability of pluton geometry; emplacement of additional magma produces vertical pluton growth rather than lateral growth. Similarly,

128

ERIC HORSMAN ET AL.

same place (along the margins) during growth. Thus, the vertical growth began after lateral growth, and consequently vertical growth was not accompanied by signiﬁcant lateral growth. Secondly, in the marginal wallrock around the Black Mesa intrusion, not only sub-vertical, but also sub-horizontal and shallowly dipping fractures and faults have been measured, with normal and reverse slip sense (Saint Blanquat et al. 2006). These can be interpreted as remnants of ﬂexural hinges developed during the beginning of the vertical growth, before the loss of continuity of the bedding (see ﬁg. 16 of Corry 1988). The question remains as to what factors control the change from sill-like emplacement to bysmalith-like emplacement during the growth of a large intrusion. As stated above, the present authors’ observations contradict Pollard & Johnson’s (1973) suggestion that the switch occurs once a laccolith reaches a particular radius relative to its thickness. Another possible controlling factor may be wallrock lithology, as suggested by the emplacement of the Black Mesa intrusion within relatively compliant shale (the Summerville formation), whereas the two smaller intrusions both formed within relatively stiﬀ sandstone (the Entrada formation). Additional possibilities include an increase in magma emplacement rate, or variations in feeder position and geometry.

6.2. Fabric and magma flow patterns Figure 9 Schematic cross-sections through the Maiden Creek, Trachyte Mesa, and Black Mesa intrusions to illustrate idealised emplacement mechanisms and our preferred assembly histories. Intrusions are not drawn to scale. The number of pulses shown is intended to demonstrate relationships between successive pulses rather than actual number.

the cylindrical shape of the Black Mesa intrusion may have evolved from a less volumetrically eﬃcient elongate laccolith, similar to the Trachyte Mesa intrusion, at an earlier stage. 6.1.1. Comparison with previous models. Pollard & Johnson (1973) developed a model of the evolution of pluton geometry that was mainly based on wallrock geometry. This model involves the following three stages: (1) Intrusion of an initial sill and growth by lateral propagation; (2) When a critical length/thickness ratio is achieved, the sill begins to inﬂate vertically by overburden bending to produce a laccolith, which is also still growing laterally; (3) Finally, fracturing of wallrocks over the periphery of the intrusion permits the uplift of both the wallrock and older igneous sheets. This faultcontrolled roof uplift results in vertical growth and transforms the laccolith into a bysmalith. The observations of the present authors are in general agreement with this succession of events (sill to laccolith to bysmalith), but not in agreement with the proposed mechanisms. First, no clear evidence is found for a true laccolithic stage of growth, for the three following reasons: (1) Each intrusion has a tabular, nearly ﬂat roof, not a dome-like shape predicted by laccolithic growth; (2) Proﬁles of the contacts on the margins of all three intrusions (including the W side of Black Mesa) have a staircase-like shape rather than a smooth progressive bending predicted by classic laccolith models; and (3) The existence of simultaneous vertical and lateral propagation is not supported by the data. The initial sill of the Trachyte Mesa laccolith, for example, had a size similar to the ﬁnal horizontal size of the pluton. The idea of the achievement of the maximum horizontal size before the beginning of vertical growth is also supported by structural data from the wallrock. First, the absence of faults (i.e. ‘remnant’ hinges) in wallrock above all the intrusions shows that deformation remained localised more or less at the

The rapid cooling of the intrusions due to their small size and shallow depth ensures that the fabrics preserved in the plutons reﬂect magma ﬂow patterns. Thus, by combining detailed knowledge of intrusion geometry with fabric data from the intrusions, the evolution of magma ﬂow during emplacement and assembly can be considered. Additionally, the observations of feeder conduit geometry allow the examination of relationships between magma ﬂow patterns and the magma supply. Bulk magma ﬂow patterns evolve as intrusion geometry progresses from sill to laccolith to bysmalith. In the Maiden Creek sill, a direct correlation is observed between the complex, lobate intrusion geometry and fabric patterns. Magma ﬂowed radially out from an unexposed feeder into the main body and then was channelised into the lobate ﬁnger-like regions. The observed correspondence between geometry and fabric pattern thus provides a useful test of the utility of fabric analysis in these rocks. These complex, lobate geometries are not conﬁned to the upper crust; Stevenson et al. (2007) inferred assembly of a mid-crustal pluton through emplacement of numerous ﬁnger-like magma pulses. Magma ﬂow patterns in the Trachyte Mesa intrusion reﬂect more organisation of the magma distribution system than developed in the Maiden Creek intrusion. Rather than ﬂowing radially away from the feeder source in numerous sinuous lobes, magma ﬂowed from the several sub-horizontal feeder pipes and, once in the intrusion, out along the central axial magma channel. From this channel, magma ﬂowed out laterally to supply the margins of the intrusion. This bilaterally symmetric magma ﬂow pattern bears a striking resemblance to inferred magma ﬂow patterns in large, complex sills analysed in detailed three-dimensional seismic reﬂection studies (Thomson & Hutton 2004; Hansen & Cartwright 2006) and to physical models of magma ﬂow by Kratinova´ et al. (2006). Although these patterns are commonly interpreted as resulting from ﬂow away from a steeply dipping feeder dike (e.g. Ferre´ et al. 2002), these examples demonstrate that feeder geometry cannot be inferred from fabric data alone. Magma ﬂow patterns in the Black Mesa intrusion have a bilateral symmetry similar to that observed in the Trachyte Mesa intrusion. Both intrusions have central axial conduits

EMPLACEMENT AND ASSEMBLY OF SHALLOW INTRUSIONS

trending radially away from the main Mt Hillers intrusive centre. Additionally, the lateral margins of the both intrusions were presumably fed by ﬂow away from an axial conduit. This similarity suggests that the magma distribution system may be similar in both the laccolith and the bysmalith. However, little fabric data from the interior of the Black Mesa intrusion is available, so this hypothesis is not well constrained. Nonetheless, the similarity suggests that early in its assembly history (i.e. preserved in fabrics near the roof) the Black Mesa intrusion went through a stage similar to that observed on the Trachyte Mesa intrusion. By extension, if the transition from laccolith to bysmalith involved vertical inﬂation and progression toward a cylindrical shape, it is consistent with the magma distribution system remaining stable throughout that geometric evolution. Thus, it appears that with increasing maturity of the magma distribution system, ﬂow patterns evolve and become progressively more stable and eﬃcient. This analysis of correspondence between evolving intrusion geometry and magma ﬂow patterns was possible only because the conditions were ideal for preservation of fabric in these small intrusions.

6.3. Pulsed assembly Evidence of pulsed assembly exists in each of the three intrusions. This evidence is unequivocal in the Maiden Creek and Trachyte Mesa intrusions. In the Black Mesa intrusion, the sub-horizontal textural layering and cataclastic zones near the top of the intrusion suggest that early stages of assembly involved multiple magma pulses. The importance of magma pulses during the late stages of assembly of the Black Mesa intrusion is less clear, but vertical variation in geochemistry suggests that pulses existed throughout assembly (Bankuti 2007). Assuming that pulsed assembly continued throughout the growth of the intrusion, a threshold governing sheet contact preservation was crossed at some point. This threshold is of considerable interest because of the ongoing debate regarding possible pulsed assembly of large intrusions. Several factors govern the preservation of evidence for pulses. Diﬀerences in composition of juxtaposed pulses result in the most easily recognisable contacts. A large temperature contrast between the pulse and the host rock (i.e. wallrock or a previous magma pulse) favours the development of solidstate fabric and makes pulse contacts more easily recognisable. The strain rate in the intruding magma must however be suﬃciently high to produce solid-state fabric. A long time-lag between pulses allows the previous pulse to cool and favours strong fabric development. These several factors are interdependent and govern processes like annealing of emplacement-related fabric and reworking of pulse margins, both of which can destroy evidence of pulses. For example, Tuﬀen & Dingwell (2005) demonstrate how early cataclastic fabrics in shallow volcanic conduits were modiﬁed and eventually destroyed by subsequent annealing and ductile deformation. They note that in some cases it is possible to recognise multiple generations of such overprinting structures. This cyclical brittle–ductile deformation may be a consequence of temperature variations moving the melt across liquid–glass transition. Thus, a new pulse of magma may reheat older solidiﬁed magma, including solid-state fabric, resulting in annealing and renewed ﬂow deformation of the older magma. Matzal et al. (2006) present further evidence of reworking of pulse margins deeper in the crust. In their model, as a magma pulse is emplaced and begins to cool, the volumetric crystal percentage increases, and the remaining melt is largely trapped inside a crystal-rich carapace. Subsequent pulse emplacement reheats the margin of the previous sheet, partially incorporating some of the other material, and blur-

129

ring the transition between the pulses both chemically and texturally. The question remains at to why sheeting is clearly preserved in the Maiden Creek and Trachyte Mesa intrusions, but cryptic in the Black Mesa intrusion. Both the shallow depth of emplacement and the relatively small volume of all three intrusions contribute to rapid cooling. The lack of preservation in the Black Mesa intrusion demonstrates the ephemeral nature of such features, even with rapid cooling. The lack of ﬁeld evidence of pulsed assembly in some deeper, larger intrusions (e.g. Glazner et al. 2004) is therefore not surprising. Pulsed assembly is, however, apparent in some plutons, despite the factors working against preservation of evidence. Most examples of large intrusions with pulsed assembly are composite plutons where pulses of diﬀerent composition make recognition straightforward (Wiebe 1993, 1994; Wiebe & Collins 1998; Wiebe et al. 2002; Matzal et al. 2006; Michel et al. 2008), although evidence can be cryptic even in composite plutons (Mahan et al. 2003). Host rock deformation can provide indirect evidence of magma pulses. Albertz et al. (2005) suggest that moderate background strain rates in pluton aureoles may be punctuated by high strain rate excursions associated with emplacement of magma pulses. Ferna´ndez & Castro (1999) called upon similar processes. These pulses of intense deformation can produce localisation of strain as shear zones or faults in the vicinity of plutons. This intense deformation is superposed upon the background deformation of host rock associated with pluton emplacement. Several hypotheses exist for the formation of magma pulses and can be broadly divided into mechanisms due to: (1) ascent processes; (2) emplacement processes; or (3) source region processes. Ascent-related mechanisms suggest that interaction between magma ascending in sub-vertical dike-like or pipe-like channels results in formation of magma pulses (e.g. Denlinger & Hoblitt 1999). Emplacement-related mechanisms attribute magma pulses to similar interactions during pluton assembly rather than in the feeder conduit. The processes of magma generation and segregation in the source region operate an order of magnitude more slowly than the processes of ascent and emplacement (Petford et al. 2000). Thus, even minor cyclical processes in the source region can result in more dramatic cyclicity in the upper crust. Regardless of the mechanisms leading to pulse formation, the existence of pulses may largely invalidate existing mechanical models of the growth of shallow intrusions (e.g. Pollard & Johnson 1973; Kerr & Pollard 1998; Zenzri & Keer 2001), which rely on entire igneous bodies acting as a single mechanical entity throughout assembly. Future theoretical models should account for empirical evidence for pulsed assembly and actual geometric shapes (e.g. ﬂat rather than domed tops) of the intrusions. Further ﬁeld-based work needs to concentrate on how magma ﬂows during a single pulse, which can in turn constrain mechanical models of emplacement and assembly.

6.4. Time scales of emplacement and assembly Recognition of the pulsed nature of the assembly of many upper-crustal plutons leads to the question of the time scales on which these processes operate. Few timing constraints exist for the Henry Mountains, but radiometric ages constrain timing at the scale of the entire range and, to a lesser extent, in individual intrusive centres (Nelson et al. 1992). For example, ages throughout the Henry Mountains span w9 Myr from 30 Ma to 21 Ma. Individual intrusive centres have age ranges restricted to w3 Myr or less. Other data provide evidence of the relative ages of component intrusions in intrusive centres. For example, Jackson & Pollard (1988) used palaeomagnetism

130

ERIC HORSMAN ET AL.

to demonstrate that early sills on Mt Hillers were rotated and uplifted by later magmatism that formed the underlying main laccolithic body. Thermal modelling can provide additional constraints on assembly times for individual plutons (e.g. Annen et al. 2006). The present authors use one-dimensional thermal modelling of the Black Mesa intrusion by Habert & Saint Blanquat (2004) and Saint Blanquat et al. (2006) to provide a conservative maximum duration of pulsed assembly. Several assumptions and observations ensure that this thermal modelling provides a maximum estimate of the duration of assembly. First, the assumption of cooling solely through conduction provides a maximum estimate of the time necessary to cool through the solidus, because the advection of heat due to interaction with groundwater, or to the release of intrusion-related ﬂuids, would cool the intrusion more rapidly than conduction alone. Secondly, the lack of evidence of clear contacts between separate magma pulses in the pluton constrains the time between pulses to be no longer than the time for the margin of last pulse to cool through the solidus. Results of the thermal model suggest that the lack of internal contacts can be explained if the intrusion formed in less than 100 years through the amalgamation of numerous sheets. This result implies that the roof lifted at least 2 m/yr during assembly. The less voluminous Maiden Creek and Trachyte Mesa intrusions were likely assembled in much shorter periods of time. The time scales of the emplacement of individual component magma pulses are diﬃcult to assess directly. However, individual pulses in each intrusion must be short lived, particularly if individual plutons are emplaced within 100 years (e.g. the Black Mesa intrusion). The magma supply clearly persisted for a lengthy period of time relative to the lifespan of a single pluton, or even a single intrusive centre. At the scale of the Mt Hillers intrusive centre, magma may have ascended in a few relatively large batches. However, the satellite intrusions studied in the present paper were likely assembled in myriad small, laterally fed pulses from the adjacent larger body. Evidence from other studies reinforces the observations made in the Henry Mountains. Examples of the pulsed nature of assembly come from studies of both modern and ancient upper-crustal magmatism. Interferometric synthetic aperture radar studies demonstrate that sub-volcanic magma chambers inﬂate and deﬂate on time scales of weeks to months (e.g. Voight et al. 1998; Pritchard & Simons 2004). Henry et al. (1997) note that the Solitario laccolith-caldera is composed of three distinct episodes of magmatism, and demonstrate that these episodes are each composed of numerous magma pulses. Evans et al. (1993) used seismic reﬂection to recognise that the Lake District batholith is composed of numerous subhorizontal sheets. Similarly, radiometric dating of such intrusions suggests that magmatism feeding a single intrusion can last a million years or even a few million years (Harrison et al. 1999; Glazner et al. 2004; Matzal et al. 2006). Such time spans are considerably longer (often by an order of magnitude) than it takes a shallow pluton assembled as a single large pulse to solidify. Thus, several independent lines of evidence suggest that pluton assembly in the upper crust is not a continuous and gradual process. Rather, many plutons grow through the amalgamation of numerous pulses of magma. These pulses tend to be obscured even under conditions ideal for preserving evidence of magma pulses: high crustal levels, fast cooling rates, and no regional deformation. The characterisation of pulses likely depends on both the temporal and the spatial scales of observation.

6.5. Emplacement versus assembly The exceptional exposure and ideal conditions for preserving evidence of magma pulses in the Henry Mountains allow us to make some statements about emplacement and assembly of plutons that are generally applicable to intrusions. The ﬁrst concerns the diﬀerence between emplacement and assembly. The intrusions of the Henry Mountains inspired Gilbert (1877) to coin the term laccolite, which he used to describe a pluton that made space for itself by uplifting the overlying rocks. For the last 100 years, there has been no argument about the emplacement of the small intrusions in the Henry Mountains. The assembly of these plutons, in contrast, has not been understood. Although previously inferred to have formed from a single pulse of magma, the present work demonstrates that they occurred in multiple magma pulses. Thus, emplacement and assembly can be distinguished. The process of assembly by multiple pulses opens new ﬁelds of research. For instance, how does an individual magma pulse ﬂow either into a pre-existing chamber or into wall rock? How do these patterns and processes evolve as subsequent pulses are emplaced? At present, these processes are largely unstudied. Another unique aspect of the Henry Mountains is the well-exposed intrusion geometries, which allow the present authors to recognise that none of the three intrusions studied ﬁts neatly into a common idealised geometry category. The Maiden Creek sill has a highly complex map-view geometry, despite its simple cross-sectional geometry. The Trachyte Mesa laccolith is largely tabular, similar to a sill, only bending the wallrocks on the margins of the intrusion. This pattern is consistent with most laccoliths, which commonly have ﬂat rather than domed tops (Corry 1988). The Black Mesa intrusion is clearly transitional between a laccolith (on the west side) and a bysmalith (on the east side). These details of true intrusion geometry provide essential information about emplacement and assembly history. While idealised models provide a useful reference frame, they can lead to oversimpliﬁcation of the mechanisms of emplacement, assembly, and magma ﬂow. Finally, there is the historical aspect of pluton emplacement and assembly. Many plutons assembled from multiple pulses change geometry considerably over time, as emplacement mechanisms accommodating sequential pulses evolve (e.g. McNulty et al. 2000; Belcher & Kisters 2006). The complexities inherent in analysing plutons in tectonically active regions (e.g. overprinting of fabrics, incomplete data sets, unknown threedimensional geometries, etc.) make it diﬃcult to directly compare them with the Henry Mountains intrusions. However, many of the same features are found in both cases, suggesting that similar processes operate during emplacement and assembly, regardless of diﬀerences in scale or the presence of tectonic activity (e.g. McCaﬀrey & Petford 1997; Saint Blanquat et al. in press).

7. Conclusions The three intrusions studied in the Henry Mountains represent successive snapshots of the emplacement of an upper crustal pluton during geometric evolution from sill to laccolith to a piston-like bysmalith. Assembly of each of these intrusive bodies occurred through a series of magma pulses, manifested locally as distinct magma sheets. The exceptionally wellpreserved intrusion geometries demonstrate that idealised models of intrusion shape are overly simplistic. These bodies appear to be fed by sub-horizontal, pipe-like conduits emanating from the main intrusive centre, located several kilometres away.

EMPLACEMENT AND ASSEMBLY OF SHALLOW INTRUSIONS

Magma ﬂow patterns are well preserved because of rapid cooling due to the small intrusion volumes, rapid assembly, and shallow depth of emplacement. These magma ﬂow patterns correspond well with observed complex intrusion geometries. Early magma ﬂow in the sill involved radial ﬂow away from a pipe-like feeder and into several ﬁnger-like lobes. As intrusion volume grew, the magma distribution system stabilised through the formation of a central axial conduit that fed magma laterally out to the pluton margins. During further growth this magma distribution system remained stable, and consistent magma ﬂow patterns reﬂect this stability. Increasing intrusion volume also resulted in the evolution of the primary emplacement mechanism: from wallrock uplift; through rotation; to distortion at intrusion margins; to wallrock uplift through faulting. Evidence of the magma pulses comprising each body becomes more cryptic as intrusion volume grows. Many processes can obscure evidence of pulses and these processes typically become more eﬀective at deeper crustal levels. Thus, a lack of evidence does not necessarily imply a lack of pulsed assembly. It is concluded that assembly of these igneous intrusions occurs through the amalgamation of numerous pulses of magma, which may be a common occurrence in upper crustal intrusions. These pulsed emplacement and assembly processes operate at many spatial and temporal scales. The conditions in the Henry Mountains – shallow crustal levels, fast cooling rates, and no regional deformation – were ideal for the preservation of internal structures. As the conditions of emplacement and assembly for most plutons are considerably more complex, preservation of primary internal structures is less likely.

8. Acknowledgements We thank participants from the 2005 G.S.A. ﬁeld trip to observe these intrusions for numerous thoughtful discussions and an abundance of constructive criticism. Buzz Rackow at the Hanksville B.L.M. oﬃce and Dave Dilloway provided logistical assistance. Sandy Cruden and an anonymous reviewer helped improve the manuscript substantially. This work was funded by NSF grant EAR-0003574 to BT and SM. Additional funding came from CNRS/NSF grant 12971 to MSB, NSF grant EAR-0510893 grant to BT, a FRCE-CMU grant to SM, and a Shell Undergraduate Fellowship at the University of Wisconsin to RAH.

9. References Albertz, M., Paterson, S. R. & Okaya, D. 2005. Fast strain rates during pluton emplacement: magmatically folded leucocratic dikes in aureoles of the Mount Stuart Batholith, Washington, and the Tuolumne Intrusive Suite, California. Geological Society of America Bulletin 117, 450–65. Annen, C., Scaillet, B. & Sparks, R. S. J. 2006. Thermal constraints on the emplacement rate of a large intrusive complex: the Manaslu Leucogranite, Nepal Himalaya. Journal of Petrology 47, 71–95. Bankuti, M. 2007. Pe´tro-ge´ne´alogie des intrusions du Mont Hillers, Utah, USA. Unpublished M.S. Thesis, Universite´ Paul-Sabatier, Toulouse, France. 57 pp. Belcher, R. W. & Kisters, A. F. M. 2006. Syntectonic emplacement and deformation of the Heerenveen Batholith; conjectures on the structural setting of the 3·1 Ga granite magmatism in the Barberton granite-greenstone terrain, South Africa. Geological Society of America, Special Paper 405, 211–32. Coleman, D. S., Gray, W. & Glazner, A. F. 2004. Rethinking the emplacement and evolution of zoned plutons; geochronologic evidence for incremental assembly of the Tuolumne Intrusive Suite, California. Geology 32, 433–6. Corry, C. E. 1988. Laccoliths: Mechanics of emplacement and growth. Geological Society of America, Special Paper 220, 110 pp.

131

Denlinger, R. P. & Hoblitt, R. P. 1999. Cyclic eruptive behavior of silicic volcanoes. Geology 27, 459–62. Fernandez, C. & Castro, A. 1999. Pluton accommodation at high strain rates in the upper continental crust; the example of the Central Extremadura Batholith, Spain. Journal of Structural Geology 21, 1143–9. Evans, D. J., Rowley, W. J., Chadwick, R. A. & Millward, D. 1993. Seismic reﬂections from within the Lake District Batholith, Cumbria, Northern England. Journal of the Geological Society, London 150, 1043–6. Ferre´, E. C., Bordarier, C. & Marsh, J. S. 2002. Magma ﬂow inferred from AMS fabrics in a layered maﬁc sill, Insizwa, South Africa. Tectonophysics 354, 1–23. Gaﬀney, E. S., Damjanac, B. & Valentine, G. A. 2007. Localization of volcanic activity: 2. Eﬀects of pre-existing structure. Earth and Planetary Science Letters 263, 323–8. Gilbert, G. K. 1877. Report on the Geology of the Henry Mountains. U.S. Geological Survey. 170 pp. Glazner, A. F., Bartley, J. M., Coleman, D. S., Gray, W. & Taylor, R. Z. 2004. Are plutons assembled over millions of years by amalgamation from small magma chambers? GSA Today 14, 4–11. Habert, G. & de Saint-Blanquat, M. 2004. Rate of construction of the Black Mesa bysmalith, Henry Mountains, Utah. In Breitkreuz, C. & Petford, N. (eds) Physical Geology of High-level Magmatic Systems. Geological Society, London, Special Publication 234, 163–730. Bath, UK: The Geological Society Publishing House. Hansen, D. M. & Cartwright, J. 2006. Saucer-shaped sills with lobate morphology revealed by three-dimensional seismic data: implications for resolving a shallow-level sill emplacement mechanism. Journal of the Geological Society, London 163, 509–23. Harrison, T. M., Grove, M., Lovera, O. M., Catlos, E. J., D’Andrea, J. & McKeegan, K. 1999. Origin and episodic emplacement of the Manaslu Intrusive Complex, Central Himalaya. Journal of Petrology 40, 3–19. Henry, C. D., Kunk, M. J., Muehlberger, W. R. & McIntosh, W. C. 1997. Igneous evolution of a complex laccolith-caldera, the Solitario, Trans-Pecos Texas: implications for calderas and subjacent plutons. Geological Society of America Bulletin 109, 1036–54. Horsman, E., Tikoﬀ, B. & Morgan, S. 2005. Emplacement-related fabric and multiple sheets in the Maiden Creek Sill, Henry Mountains, Utah, USA. Journal of Structural Geology 27, 1426–44. Hunt, C. 1953. Geology and geography of the Henry Mountains region, Utah. US Geological Survey Professional Paper 228. Hutton, D. H. W. 1988. Granite emplacement mechanisms and tectonic controls; inferences from deformation studies. Transactions of the Royal Society of Edinburgh: Earth Sciences 79, 245–55. Hutton, D. H. W. 1997. Syntectonic granites and the principle of eﬀective stress. In Bouchez, J. L., Hutton, D. H. & Stephans, W. E. (eds) Granite: From segregation of melt to emplacement fabrics, 189–98. Dordrecht: Kluwer. Jackson, M. D. & Pollard, D. D. 1988. The laccolith-stock controversy: new results from the southern Henry Mountains, Utah. Geological Society of America Bulletin 100, 117–39. Jackson, M. D. & Pollard, D. D. 1990. Flexure and faulting of sedimentary host rocks during growth of igneous domes, Henry Mountains, Utah. Journal of Structural Geology 12, 185–206. Johnson, A. M. & Pollard, D. D. 1973. Mechanics of growth of some laccolithic intrusions in the Henry Mts, Utah, I: Field observations, Gilbert’s model, physical properties and the ﬂow of magma. Tectonophysics 18, 261–304. Kerr, A. D. & Pollard, D. D. 1998. Toward more realistic formulations for the analysis of laccoliths. Journal of Structural Geology 20, 1783–93. Kratinova´, Z., Zavada, P., Hrouda, F. & Schulmann, K. 2006. Non-scaled analogue modelling of AMS development during viscous ﬂow: A simulation on diapir-like structures. Tectonophysics 418, 51–61. Lipman, P. W. 1987. The roots of ash-ﬂow calderas in western North America: windows into the tops of granitic batholiths. Journal of Geophysical Research 89, 8801–41. Lipman, P. W. 2007. Incremental assembly and prolonged consolidation of Cordilleran magma chambers: evidence from the Southern Rocky Mountain volcanic ﬁeld. Geosphere 3, 42–70. Mahan, K. H., Bartley, J. M., Coleman, D. S., Glazner, A. F. & Carl, B. 2003. Sheeted intrusion of the synkinematic McDoogle pluton, Sierra Nevada, California. Geological Society of America Bulletin 115, 1570–82.

132

ERIC HORSMAN ET AL.

Matzel, J. E. P., Bowring, S. A. & Miller, R. B. 2006. Time scales of pluton construction at diﬀering crustal levels; examples from the Mount Stuart and Tenpeak Intrusions, north Cascades, Washington. Geological Society of America Bulletin 118, 1412–30. McCaﬀrey, K. J. W. & Petford, N. 1997. Are granitic intrusions scale invariant? Journal of the Geological Society, London 154, 1–4. McNulty, B. A., Tobisch, O. T., Cruden, A. R. & Gilder, S. 2000. Multistage emplacement of the Mount Givens Pluton, central Sierra Nevada Batholith, California. Geological Society of America Bulletin 112, 119–35. Michel, J., Baumgartner, L., Putlitz, B., Schaltegger, U. & Ovthcarova, M. 2008. Incremental growth of the Patagonian Torres del Paine laccolith over 90 k.y. Geology 36, 459–62. Morgan, S. S., Stanik, A., Horsman, E., Tikoﬀ, B., de Saint-Blanquat, M. & Habert, G. 2008. Emplacement of multiple magma sheets and wallrock deformation: Trachyte Mesa intrusion, Henry Mountains, Utah. Journal of Structural Geology 30, 491–512. Morgan, S. S., Horsman, E., Tikoﬀ, B., de Saint-Blanquat, M. & Habert, G. 2005. Sheet-like emplacement of satellite laccoliths, sills and bysmaliths of the Henry Mountains, southern Utah. Geological Society of America Field Guide 6, 283–309. Nelson, S. T., Davidson, J. P. & Sullivan, K. R. 1992. New age determinations of central Colorado Plateau laccoliths, Utah: Recognizing disturbed K–Ar systematics and re-evaluating tectonomagmatic relationships. Geological Society of America Bulletin 104, 1547–60. Nelson, S. T. 1997. Reevaluation of the Central Colorado plateau laccoliths in the light of new age determination. US Geological Survey Bulletin 2158, 37–9. Paterson, S. R., Fowler, T. K., Jr., Schmidt, K. L., Yoshinobu, A. S., Yuan, E. S. & Miller, R. B. 1998. Interpreting magmatic fabric patterns in plutons. Lithos 44, 53–82. Paterson, S. R. & Fowler, T. R., Jr. 1993. Re-examining pluton emplacement processes. Journal of Structural Geology 15, 191–206. Petford, N., Cruden, A. R., McCaﬀrey, K. J. W. & Vigneresse, J. L. 2000. Granite magma formation, transport and emplacement in the Earth’s crust. Nature 408, 669–73. Pollard, D. D. 1973. Derivation and evaluation of a mechanical model for sheet intrusions. Tectonophysics 19, 233–69. Pollard, D. D. & Johnson, A. M. 1973. Mechanics of growth of some laccolithic intrusions in the Henry Mountains, Utah, Part II: Bending and failure of overburden and sill formation. Tectonophysics 18, 311–54. Pritchard, M. E. & Simons, M. 2004. An InSAR-based survey of volcanic deformation in the Southern Andes. Geophysical Research Letters 31, L15610. doi:10.1029/2004GL020545.

Saint Blanquat, M. (de), Habert, G., Horsman, E., Morgan, S., Tikoﬀ, B., Launeau, P. & Gleizes, G. 2006. Mechanisms and duration of non-tectonically assisted magma emplacement in the upper crust, the Black Mesa pluton, Henry Mountains. Tectonophysics 428, 1–31. Saint-Blanquat, M. (de), Horsman, E., Habert, G., Law, R. D., Morgan, S., Tikoﬀ, B. & Vanderhaeghe, O. in press. Multiscale magmatic cyclicity, duration of pluton construction and the paradoxical relationship between tectonism and plutonism in continental arcs. Tectonophysics. Stevenson, C. T. E., Owens, W. H. & Hutton, D. H. W. 2007. Flow lobes in granite: the determination of magma ﬂow direction in the Trawenagh Bay granite, NW Ireland, using AMS. Geological Society of America Bulletin 119, 1368–86. Thompson, G. A. & Zoback, M. L. 1979. Regional geophysics of the Colorado Plateau. Tectonophysics 61, 149–81. Thomson, K. & Hutton, D. H. W. 2004. Geometry and growth of sill complexes: insights using 3-d seismic from the North Rockall Trough. Bulletin of Volcanology 66, 364–75. Tuﬀen, H. & Dingwell, D. B. 2005. Fault textures in volcanic conduits: evidence for seismic trigger mechanisms during silicic eruptions. Bulletin of Volcanology 67, 370–87. Voight, B., Hoblitt, R. P., Clarke, A. B., Lockhart, A., Miller, A. D., Lynch, L. & McMahon, J. 1998. Remarkable cyclic ground deformation monitored in real-time on Montserrat and its use in eruption forecasting. Geophysical Research Letters 25, 3405–8. Weinberg, R. F. 1996. Ascent mechanism of felsic magmas: news and views. Geological Society of America, Special Paper 315, 95–103. Wiebe, R. A. 1993. The Pleasant Bay layered gabbro-diorite, coastal Maine; ponding and crystallization of basaltic injections into a silicic magma chamber. Journal of Petrology 34, 461–89. Wiebe, R. A. 1994. Silicic magma chambers as traps for basaltic magmas; the Cadillac Mountain intrusive complex, Mount Desert Island, Maine. Journal of Geology 102, 423–37. Wiebe, R. A. & Collins, W. J. 1998. Depositional features and stratigraphic sections in granitic plutons: Implications for the emplacement and crystallization of granitic magma. Journal of Structural Geology 20, 1273–89. Wiebe, R. A., Blair, K. D., Hawkins, D. P. & Sabine, C. P. 2002. Maﬁc injections, in situ hybridization, and crystal accumulation in the Pyramid Peak granite, California. Geological Society of America Bulletin 114, 909–20. Zenzri, H. & Keer, L M. 2001. Mechanical analyses of the emplacement of laccoliths and lopoliths. Journal of Geophysical Research 106, 13781–92.

MS received 18 January 2008. Accepted for publication 3 June 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 133–145, 2010 (for 2009)

Four-stage building of the Cambrian Carion pluton (Madagascar) M. O. M. Razanatseheno1, A. Ne´de´lec2, M. Rakotondrazafy1, J. G. Meert3 and B. Ralison1 1

Department of Earth Sciences, BP 906, Antananarivo University, Madagascar

2

LMTG – OMP, Toulouse University – CNRS-IRD, 14 avenue Edouard Belin, 31400 Toulouse, France Email: [email protected]

3

Department of Geological Sciences, 241 Williamson Hall, Gainesville University, Florida 32611, USA

ABSTRACT: The 5325 Ma old Carion pluton is a dark, porphyritic ferro-potassic granitoid emplaced near the late Pan-African Angavo mega-shear zone. A rough normal zoning from tonalitic to granitic compositions can be recognised in the ﬁeld. Steep magmatic foliations are evidenced by K-feldspar megacryst preferred orientations. Microstructures are either magmatic or typical of incipient solid-state deformation in near solidus conditions. Magnetic susceptibility magnitudes (K) range from 11 to 11110 3 SI in the pluton and can be correlated to the petrography (highest K values in the tonalites; lowest K in the granites; granodiorites in between). The susceptibility magnitudes display a complex zoning pattern. Combined with the arrangement of magnetic foliation trajectories, it is possible to delineate four nested sub-units, regarded as four magmatic pulses successively emplaced from the west to the east of the pluton. The four pulses are characterised by very similar magma geochemistry, but variable magmatic diﬀerentiation. The highest degrees of magnetic susceptibility anisotropies (up to 1·6) are observed along internal contacts between sub-units and along the borders of the pluton. The magnetic lineations are also steeply plunging in some places in each sub-unit, possibly imaging the diﬀerent feeder zones. Magma emplacement occurred at the end of the activity of the Angavo shear zone, hence avoiding re-orientation of the magmatic structures by the late Pan-African transcurrent tectonics. The diachronicity of the four magmatic pulses is consistent with previously determined palaeomagnetic data, because only the two older sub-units display a magnetic reversal sequence, whereas the two youngest sub-units lack any reversion. Emplacement of these four magmatic batches was responsible for a strain aureole and suggests a diapiric mode of ascent. KEY WORDS: pluton

anisotropy of magnetic susceptibility, diapir, granitoid, magnetic reversal, zoned

Granitic plutonism is particularly active in collision orogens in which both the thermal conditions necessary for magma genesis and the loci for magma ascent and emplacement in the crust are present. In such collisional settings, granite magma production follows continental suturing by a few tens of Mys and remains often voluminous until the last stages of orogeny (Thompson & Connolly 1995). In a pre-drift reconstruction of Gondwana (Fig. 1a), Madagascar is located in the East African Orogen, a collisional orogen formed by the closure of the Mozambique Ocean at the end of Neoproterozoic times (Stern 1994; Meert 2003). A number of granitoid plutons were emplaced in Madagascar at various stages. The late-PanAfrican history of Madagascar is dominated by transcurrent tectonics along major shear zones, especially in the southern and eastern parts of the island (Martelat et al. 1997; Ralison & Ne´de´lec 1997; Ne´de´lec et al. 2000). This present study focuses on the Carion pluton (Fig. 1b, c), located near the Angavo shear zone, a lithospheric-scale transcurrent shear zone (Ralison & Ne´de´lec 1997), that has been active at around 550 Ma, as can be inferred from the ages of spatially-related syntectonic and deformed granitic rocks (Kro¨ner et al. 1999; Gre´goire et al. 2008). The Angavo shear zone is associated with low-pressure granulitic parageneses and

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016156

prograde charnockitisation, whose P and peak T conditions were calculated at 790 (10)(C and 330 (30) MPa by Ne´de´lec et al. (2000). The Carion pluton itself crystallised at about 320 (10) MPa (i.e. at about 10 km depth) after Al-in-hornblende barometric calculations using the calibration of Schmidt (1992). It has been dated at 532 (5) Ma by SHRIMP zircon age (Meert et al. 2001b). Hence, it was slightly younger than the transcurrent tectonics along the Angavo shear zone, therefore representing one of the latest stage of Pan-African magmatism in Madagascar. On the geological map of Besairie (1969), it is classiﬁed as a post-tectonic pluton. Since the pioneer work of Ellwood & Whitney (1980), anisotropy of magnetic susceptibility (AMS) has been proven the most eﬃcient tool in determining the internal fabric of granitic plutons, as recently reviewed by Bouchez (1997; 2000 and references therein). This method is here applied to the Carion pluton. Preliminary AMS results were already provided by Ne´de´lec et al. (2000). These authors observed welldeﬁned magnetic/magmatic foliations and steeply plunging magnetic/magmatic lineations, in sharp contrast with the subhorizontal N–S lineations of the nearby Angavo shear zone, hence conﬁrming the late- to post-tectonic character of the pluton. The present paper provides a more detailed AMS study

134

M. O. M. RAZANATSEHENO ET AL.

Figure 1 (a) Gondwana reconstruction at 550 Ma after Lawver et al. (1998). (b) The East African Orogen at 550 Ma after Abdelsalam & Stern (1996), Ne´de´lec et al. (2000) and Reeves & de Wit (2001). Shear zones: A=Achankovil: BR=Bongolava-Ranostsara: M=Moyar; PC=Palghat–Cauvery: E.L.=Enderby Land. (c) Location of AMS sampling sites.

of the pluton and its wall rocks. Combined with ﬁeld geology and petrology data, it reveals the internal structure of the pluton. These data are in turn used to infer the magma emplacement mode. In addition, they highlight previous observations related to former palaeomagnetic results and Ar–Ar geochronology of Meert et al. (2001b).

1. Field and petrology data The Carion pluton extends over w370 km2 and its elliptical shape is deeply indented to the north (Fig. 1c). Country rocks are made of migmatites, pink biotite granites, hornblende gneisses and charnockites, that all belong to the Ambatolampy Group of the Graphite System (Besairie 1969; Hottin 1976). On the maps, foliations of the country rocks were drawn parallel with the massif contours, except in the northern indentation, where they seem to be crosscut by the intrusion, hence the post-tectonic setting ﬁrst proposed by Lautel (1953). The pluton area is rather hilly (Fig. 2a), with good outcrops (some of them used as quarries), due to the fact that the granitic rocks are less weatherable than their gneissic host rocks. The Carion pluton does not show any contact metamorphism at its margins, evidencing the lack of thermal contrast

between the magma and its host rocks. When observed, contacts are not sharp. There is often a transition zone up to 1 km in width, where the granitic magma formed conformable sheets within the country rocks. To the northwest, the magma was intruded as a sinuous veining network in the nearby migmatites (Fig. 2e), suggesting that the country rocks were still hot and soft. Rafts and xenoliths of migmatites of various sizes can be observed in the southern margin of the pluton (Fig. 2d). Hence, the pluton contours drawn in the maps represent a simpliﬁcation. The ﬁrst detailed petrographic account of the Carion Massif was given by Hottin (1975). This author underlined the foliated nature of the granite, but failed to recognise the mainly magmatic nature of the foliation, and, for this reason, regarded the intrusion as an orthogneissiﬁed body, a contention that is no longer tenable in the light of microscopic observations (Fig. 3). As observed by Hottin (1975), the main rock type is a mesocratic (medium to dark grey) porphyritic granitoid containing K-feldspar megacrysts with an average length of 3 cm. The feldspar preferred orientation is responsible for the conspicuous magmatic foliation observed in the ﬁeld (Fig. 2b). Apart from the megacrysts, rocks are mediumto coarse-grained, with an average grain-size of about 5 mm.

FOUR-STAGE BUILDING OF THE CAMBRIAN CARION PLUTON

135

Figure 2 (a) Characteristic dome outcrop in Carion pluton. (b) Typical porphyritic Carion granitoid with conspicuous foliation. (c) Contact between prophyritic grey granitoid and pink granite at site DG52. (d) Elongate xenolith of migmatite at site DG 5. (e) Veining network of porphyritic granitoid in wall rock migmatite at site DG 9. (f) Microgranumlar maﬁc enclave within a granitic host at site DG 5.

Scarce maﬁc microgranular enclaves are globular in shape and no more than a few tens of centimetres in length. Feldspar megacrysts were mechanically introduced from the host granitic magma into these enclaves before their complete crystallisation, pointing to their co-magmatic nature (Fig. 2f). Towards the margins of the intrusion, feldspar megacrysts are smaller and less abundant, and plagioclase prevails over alkali feldspar, hence a tonalitic composition. In the centre of the pluton, rocks are lighter in colour (light grey to pink), due to less abundant ferro-magnesian minerals and to the pink colour of the K-feldspar megacrysts; they have a granitic composition. A pink leucogranite sometimes forms magmatic dykes or larger hectometric stocks, but its porphyritic texture

is very similar to the texture of the main petrographic type (Fig. 2c). Mineral abundances and compositions are determined in thin sections and by electron microprobe analyses. Feldspars, both as megacrysts and as interstitial grains, constitute about 60% of the rocks. The K-feldspar megacrysts are made of perthitic orthose or microcline (Fig. 3a). Plagioclase are nearly unzoned, with a usually restricted composition in the range An26–20. Quartz is present as large crystals, sometimes elongate parallel to the magmatic foliation (Fig. 3b). The grains often display chessboard patterns due to both prismatic and basal subgrain boundaries, attesting for incipient solid-state deformation in near solidus conditions. However, the lack of

136

M. O. M. RAZANATSEHENO ET AL.

Figure 3 (a) Microcline megacryst (Mi), biotite (Bi) and quartz (Q) in sample DG 62 (crossed polars). (b) Elongate quartz grain (Q), slightly perthitic K-feldspar (FK) and biotite (Bi) in sample DG60 (crossed polars). (c) Cluster of hornblende (Hb) and magnetite (Mg) along the boundary of a subautomorphic plagioclase (Pl) in sample DG 52 (crossed polars). (d) Quartz (Q) (large undeformed grain), biotite (Bi) and associated magnetite (Mg) in sample DG 60 (crossed polars).

recrystallised grains testiﬁes that no signiﬁcant deformation occurred after the solidiﬁcation of the magma. Ferromagnesian silicate minerals are hornblende, actually edenite or magnesio-hornblende after the classiﬁcation of Leake et al. (1997), and biotite. They often form clusters with accessory minerals (Fig. 3c). Biotite is ubiquitous (Fig. 3b–d), whereas hornblende is rare or absent in the most evolved (granitic) rocks. Whatever the rock type, hornblende and biotite contain a relatively high magnesium fraction (XMg=0·6–0·7) and are also characterised by a relatively high ﬂuorine content (XF=0·3–0·5), as can be seen in Table 1. Accessory minerals are ilmenite, magnetite, titanite, apatite and zircon. The iron oxide compositions were determined by electron microprobe analysis (Table 1). Ilmenite is the most abundant accessory mineral and forms aggregates, together with titanite and apatite, in the interstitial spaces surrounding the large feldspar and quartz grains (Fig. 3a, c). Ilmenite crystals sometimes contain a few haematite exsolution lamellae. Magnetite is less abundant than ilmenite and is also observed as inclusions in feldspars and ferro-magnesian silicates, pointing to its earlier crystallisation in the magma. Whole-rock compositions of seven representative samples are given in Meert et al. (2001a). The darkest tonalites from the eastern and western sides have silica contents in the 61–63% range and total iron oxide contents of about 6%, whereas the most evolved rock, a pink granite from the southern central area of the pluton, has a silica content of about 70% and an iron oxide content of 2·6%. Other samples display major element contents in between these endmembers. All analysed rocks are metaluminous and follow the sub-alkaline magmatic trend in the R1–R2 diagram of La Roche et al. 1980 (Fig. 4a).

More precisely, they are highly potassic (shoshonitic) and ferriferous magmatic rocks (Meert et al. 2001a; Rakotondrazafy et al. 2001). On the basis of ﬁeld and geochemical data, the Carion pluton had been regarded so far as a simple normally zoned pluton (Fig. 4b).

2. AMS data 2.1. Material and methods Internal fabrics of granitic rocks are generally diﬃcult to measure through classic techniques based on direct observation. For instance, lineations are diﬃcult to determine in the ﬁeld, even in the case of a well foliated porphyritic granite like the Carion granite. AMS measurements provide scalar parameters as well as foliation and lineation maps that, combined with petrographic and microstructural data, make it possible to reﬁne the internal structure of the Carion pluton and to infer its emplacement history. Preliminary structural data on the Carion pluton was based on ten sampling sites in the pluton and seven sites in its wall rocks, that were part of an extensive AMS and metamorphic study in an area extending over 150 km by 70 km around Antananarivo city (Ne´de´lec et al. 2000). This ﬁrst data set was augmented by 48 sites in the plutons and 25 in the country rocks, resulting in a total of 58 sampling sites within the Carion pluton and 32 sites in the surroundings (Fig. 1c). A sampling station is characterised by two to eleven (median value: four) oriented cylinder specimens (22 mm in length and 25 mm in diameter). AMS was determined using a Kappabridge KLY 2 susceptometer (AGICO, Brno, Czech Republic) with a detection limit of about

FOUR-STAGE BUILDING OF THE CAMBRIAN CARION PLUTON

137

Table 1 Representative compositions of iron-bearing minerals. Structural formulae on the basis of 13 cations (+Ca, Na, K) for amphiboles, 22 O for biotites, three cations for magnetite and two cations for ilmenite and hematite. N=calculated; FeO*=total Fe as FeO. Amphiboles Rock type Sample no.

Biotites

Tonalites GC 24 DG 62

Granite GC 12

SiO2 TiO2 Al2O3 Fe2O3 N FeO N MnO MgO CaO Na2O K2 O F Cl Total

43·00 1·44 8·96 2·59 13·66 0·48 11·45 11·78 1·94 1·48 1·08 0·15 99·06

44·03 1·40 8·63 5·14 10·38 0·38 12·63 11·19 1·85 1·56 1·72 0·16 99·45

43·87 1·10 8·07 3·90 13·86 0·66 10·91 11·67 1·64 1·25 1·75 0·16 98·58

Si AlIV AlVI Ti Fe3+ Fe2+ Mg Mn Ca Na(B) Na(A) K Sum cations F Cl OH N

6·54 1·46 0·14 0·17 0·30 1·74 2·59 0·06 1·92 0·08 0·49 0·29 15·78 0·61 0·09 1·30

6·58 1·42 0·11 0·16 0·58 1·30 2·81 0·05 1·79 0·21 0·33 0·30 15·63 0·81 0·041 1·15

6·66 1·34 0·11 0·13 0·45 1·76 2·47 0·09 1·90 0·10 0·38 0·24 15·62 0·93 0·09 0·98

0·60 0·31

0·68 0·41

0·58 0·47

XMg XF

Magnetite Rock type Sample no. SiO2 TiO2 Al2O3 Fe2O3 N FeO MnO NiO Total Si Ti AlVI Fe3+ Fe2+ Mn2+ Ni Sum Cations

Ilmenite Tonalite MG 57 0·03 0·00 0·14 68·76 30·81 0·12 0·18 100·04 0·00 0·00 0·01 1·99 0·99 0·00 0·01 3·00

Rock type Sample no. SiO2 TiO2 Fe2O3 N FeO N MnO MgO Total Si Ti Fe3+ Fe2+ Mn2+ Mg Sum Cations Ilmenite Hematite

Rock type Sample no.

Tonalite GC 24

Granite GC 12

SiO2 TiO2 Al2O3 FeO* MnO MgO Na2O K2 O F Cl Total

37·64 3·73 13·05 14·13 0·41 15·4 0·12 10·24 2·72 0·00 0·00

37·86 2·33 12·29 15·42 0·45 14·81 0·07 9·53 2·63 0·20 0·00

Si AlIV AlVI Ti Fe Mn Mg Na K Sum cations F Cl OH N

5·68 2·32 0·00 0·42 1·78 0·05 3·46 0·04 1·97 15·74 1·27 0·00 2·73

5·77 2·23 0·00 0·41 1·96 0·06 3·36 0·02 1·85 15·68 1·23 0·05 2·85

0·66 0·32

0·63 0·30

X Mg XF

Haematite exsolved lamellae in ilmenite) Tonalite MG 57 0·03 50·20 6·59 42·67 1·47 0·57 101·53 0·00 0·94 0·12 0·89 0·03 0·02 2·00 88·63 6·16

Rock type Sample no. SiO2 TiO2 Al2O3 Fe2O3 MnO Total Si Ti Al Fe3+ Mn2+ Sum Cations

Tonalite MG 57 0·09 12·34 0·14 88·61 0·18 101·35 0·00 0·23 0·00 1·68 0·00 1·92

138

M. O. M. RAZANATSEHENO ET AL.

Figure 4 (a) Plot of Carion whole-rock compositions in the R1–R2 diagram of La Roche et al. (1980). (b) Location of analysed samples and inferred normal concentric zoning.

410 8 SI. The magnetic susceptibility magnitude (or bulk magnetic susceptibility) is the arithmetic mean deﬁned by K=1/3 (k1+k2+k3), where k1, k2 and k3 are the respective magnitudes of the principal axes of the AMS ellipsoid (Nagata 1961). The anisotropy degree is given by P=K1/K3, and the linear and planar anisotropies are L=K1/K2 and F=K2/K3 respectively. The results are presented in Table 2. Averages of magnetic foliations and lineations from each site can be seen in the projection diagrams of Figure 5. Foliations and lineations are generally well-deﬁned in most sites, with average F values (1·14) comparable to average L values (1·10). In a few cases, some scattering of data is observed, corresponding to average angular departures from the mean axis a(Ki) >5( (Table 2). This is attributed to the porphyritic nature of the rocks, with large feldspar megacrysts that can locally disturb the mineral fabric at the specimen scale.

2.2. Magnetic susceptibility magnitudes Magnetic susceptibility magnitudes (K) range from 11 to 11110 3 SI, with an average at 6210 3 SI in the Carion pluton (Table 2). These large magnetic susceptibilities are typical of magnetite-bearing granitoids. Indeed, Rochette et al. (1992) recognised that magnetite-free granitoids are characterised by K values generally less than 0·510 3 SI. The susceptibility magnitudes of Carion granitoids are in the highest observed range for granitic rocks (Bouchez 2000). Ilmenite, that is abundant in the Carion granitoids, can also have contributed eﬃciently to the susceptibility magnitudes, as this oxide has a rather high susceptibility (Borradaile & Henry 1997; Diot et al. 2003). Thermomagnetic curves (K vs. temperature) obtained from representative samples show a fast susceptibility decrease at about 580(C, the Curie temperature of pure magnetite (Fig. 6). This is evidence of the ferrimagnetic nature of these rocks, where magnetite is indeed the main susceptibility carrier in addition to ilmenite, hornblende and biotite. Outside the limits of the Carion pluton, the country rocks have generally lower susceptibilities (Fig. 7a). Within the pluton, the magnetic susceptibility map displays a pattern more complicated than the expected simple normal zoning. In paramagnetic (magnetite-free) granites, the susceptibility magnitudes are directly proportional to the iron contents of the

rocks, thus providing a reliable petrographic indicator (Gleizes et al. 1993). In ferrimagnetic rocks, the magnetic susceptibility magnitude is less straightforward to use, but it often displays some relationships with the petrography, as it mainly depends on the amounts of magnetite (in addition to other iron-bearing minerals). In intermediate to felsic metaluminous rocks such as the Carion pluton, it is expected that the amount of Fe–Ti oxides decreases as a result of magmatic diﬀerentiation. The susceptibility map conﬁrms that a diﬀerentiated area does exist in the centre of the pluton, corresponding to the pink granite type. Other minor areas with relatively low susceptibilities are also correlated to the most felsic granites. In the Carion pluton, the susceptibility values display a zoning pattern (Fig. 7a), where the lowest values correspond to the previously recognised granitic core and the highest values correspond to the tonalitic compositions near the city of Manjakandriana. However, the zoning pattern is not a simple concentric one, such as the ﬁrst pattern suggested from whole-rock geochemistry (Fig. 4b). This complex zoning pattern is explained by comparison with the foliation map (Fig. 7c). Indeed, the whole set of scalar and directional susceptibility data is consistent with the recognition of four sub-units, as discussed hereafter. The magnetic fabric of magnetite-bearing plutonic rocks has been demonstrated to parallel the shape-preferred orientation of magnetite grains (Archanjo et al. 1995; Gre´goire et al. 1998). Moreover, in the case of the Carion pluton, the consistency of the magnetic foliation with the mineral foliation due to the preferred orientations of feldspar megacrysts has been checked locally in the ﬁeld (Ne´de´lec et al. 2000). The magnetic fabric is therefore a good proxy for the mineral fabric. In the Carion pluton, foliations are rather steep and arranged in patterns that locally crosscut each other and help to delineate four nested sub-units inside the pluton (Fig. 7c). The emplacement order of the sub-units is derived from the crosscutting relationships of the foliation trajectories. The sub-units are numbered 1 to 4 from east to west according to their nesting relationships, hence their temporal relationships. The steepest dips generally coincide with external or internal contacts. Foliation trajectories are rather consistent with the limits of the pluton, with the exception of the northern central indentation, where they seem to be at a high angle to the pluton contour. In fact, in this case, foliation trajectories outside the pluton can also be

FOUR-STAGE BUILDING OF THE CAMBRIAN CARION PLUTON

139

Table 2 AMS data for Carion pluton and its country rocks. K=susceptibility magnitude; K1=magnetic lineation; K3=magnetic lineation; ‘P, L, F=anisotropy parameters; n=number of measured specimens; a(K1)=average angular departure from the mean K1 at 95% conﬁdence level; a(K3)=average angular departure from the mean K3 at 95% conﬁdence level. Sample

n

Carion pluton DG 4 2 DG 6 2 DG 16 3 DG 17 1 DG 24 4 DG 25 7 DG 26 5 DG 27 5 DG 29 4 DG 30 3 DG 31 3 DG 32 4 DG 33 5 DG 35 4 DG 36 4 DG 37 3 DG 38 4 DG 39 6 DG 40 3 DG 41 8 DG 42 3 DG 43 4 DG 44 4 DG 45 3 DG 46 4 DG 48 4 DG 49 3 DG 50 2 DG 51 4 DG 52 4 DG 53 4 DG 54 3 DG 55 4 DG 56 3 DG 57 4 DG 58 2 DG 59 4 DG 60 3 DG 61 3 DG 62 2 DG 63 3 DG 64 5 DG 65 5 DG 66 5 DG 67 3 DG 68 5 DG 69 4 DG 71 2 MG 57 6 MG 148 4 MG 149 4 TV 36 3 TV 37 3 TV 38 4 TV 41 2 TV 42 2 TV 43 4 TV 44 4 averages

K

K1

K3

(10 3 SI)

Trend

Plunge

a (K1)

Trend

Plunge

a (K3)

51 73 45 57 44 63 75 64 73 67 88 58 74 24 47 101 58 59 61 76 69 14 86 43 31 42 42 29 64 62 56 30 73 59 82 91 64 100 32 71 56 111 73 103 83 81 45 58 51 78 73 64 11 65 44 53 52 88 62

97 116 131 124 83 226 159 296 200 94 144 161 200 193 107 127 69 355 74 94 217 61 172 191 351 173 223 88 304 76 147 87 191 91 206 146 134 89 12 152 136 119 159 119 168 158 116 94 129 147 180 74 89 0 129 175 168 343

36 49 67 41 49 54 37 64 45 71 67 44 49 29 35 25 72 70 73 68 67 22 20 35 26 47 13 28 25 71 37 35 70 76 22 52 49 43 68 8 64 85 74 75 46 44 32 44 53 51 49 59 8 76 52 70 48 67

1 2 7 0 5 1 2 2 2 2 1 1 2 2 3 5 2 1 2 1 2 3 1 3 5 3 3 2 5 3 1 2 1 1 9 3 5 1 3 1 1 1 1 2 4 1 1 1 1 3 3 2 2 1 3 2 1

0 263 294 257 235 73 37 116 28 217 20 345 0 295 14 23 306 185 302 268 41 161 354 38 154 320 323 332 121 272 285 307 73 231 303 279 303 338 211 327 253 310 343 277 51 40 275 289 253 287 308 279 343 164 42 287 305 105

8 36 29 38 34 43 30 26 45 10 13 46 42 54 11 37 11 21 23 21 23 7 67 69 65 39 38 36 65 18 43 50 9 10 18 28 31 23 22 81 12 6 15 12 22 25 58 45 23 32 30 35 29 12 11 7 34 14

4 2 1 0 1 1 3 1 2 1 2 1 1 2 2 2 2 1 2 1 1 1 5 6 1 2 4 2 2 2 2 1 1 1 2 1 4 3 1 1 1 9 1 1 1 4 2 1 1 1 1 1 1 2 1 2 3

P=K1/K3

L=K1/K2

F=K2/K3

1·18 1·25 1·23 1·38 1·24 1·28 1·16 1·16 1·26 1·31 1·36 1·27 1·27 1·08 1·14 1·16 1·13 1·32 1·22 1·41 1·62 1·09 1·15 1·07 1·17 1·14 1·11 1·43 1·15 1·27 1·19 1·30 1·26 1·31 1·18 1·37 1·28 1·22 1·39 1·22 1·21 1·23 1·32 1·18 1·21 1·30 1·64 1·46 1·24 1·17 1·22 1·33 1·11 1·21 1·16 1·27 1·15 1·22 1·25

1·12 1·11 1·03 1·09 1·04 1·12 1·07 1·05 1·11 1·10 1·23 1·10 1·08 1·04 1·06 1·05 1·07 1·16 1·11 1·14 1·19 1·02 1·09 1·04 1·04 1·06 1·06 1·22 1·05 1·07 1·10 1·10 1·11 1·15 1·06 1·06 1·08 1·12 1·09 1·12 1·11 1·15 1·16 1·06 1·04 1·21 1·43 1·21 1·14 1·13 1·06 1·06 1·03 1·14 1·09 1·05 1·06 1·15 1·10

1·06 1·14 1·20 1·29 1·20 1·16 1·09 1·11 1·15 1·20 1·13 1·17 1·19 1·04 1·08 1·11 1·06 1·16 1·11 1·27 1·43 1·07 1·06 1·03 1·13 1·08 1·05 1·21 1·10 1·20 1·09 1·20 1·15 1·16 1·12 1·31 1·20 1·10 1·30 1·10 1·10 1·08 1·16 1·12 1·17 1·09 1·21 1·20 1·09 1·14 1·14 1·27 1·08 1·07 1·07 1·22 1·09 1·07 1·14

140

M. O. M. RAZANATSEHENO ET AL.

Table 2 Continued. Sample

n

Country rocks DG 1 3 DG 2 3 DG 3 3 DG 5 2 DG 7 4 DG 8 4 DG 9 3 DG 10 2 DG 12 10 DG 13 3 DG 14 2 DG 15 3 DG 18 6 DG 19 5 DG 20 6 DG 21 11 DG 22 5 DG 23 4 DG 27 11 DG 28 4 DG 30 5 DG 34 2 DG 59 4 DG 72 2 DG 73 3 MG 58 4 MG 59 4 TAN 1 3 TV 31 4 TV 32 2 TV 39 4 TV 40 4 averages

K

K1

K3

(10 3 SI)

Trend

Plunge

a (K1)

Trend

Plunge

a (K3)

52 55 9 20 26 24 35 7 7 35 18 28 9 13 13 18 18 20 58 43 46 5 17 3 37 13 6 62 18 51 30 13 25

18 357 118 77 160 213 39 53 268 329 291 350 314 157 144 156 146 145 208 341 145 54 115 217 107 219 192 358 351 350 239 339

40 67 16 7 37 23 49 12 19 17 9 37 29 8 13 13 20 15 48 45 68 27 54 8 34 20 1 37 67 32 46 66

11 1 4 2 10 2 2 2 15 2 5 6 3 2 1 2 2 7 2 4 2 2 5 1 1 4 2 1 5 2

122 177 258 212 26 75 246 295 159 192 193 177 220 287 6 269 265 246 303 75 45 254 303 126 335 320 281 269 176 191 132 162

22 29 68 80 46 58 32 66 30 67 53 53 5 78 73 60 52 39 2 12 3 63 31 2 46 32 13 3 22 64 13 20

2 1 2 1 2 3 2 2 10 1 2 1 6 5 1 1 1 2 1 1 1 1 4 2 2 1 4 2 1 1

traced inside and parallel the contact between sub-units 2 and 3. The low-susceptibility leucocratic granite core is included in sub-unit 3 that presents the widest range of susceptibilities, likely recording the largest magmatic diﬀerentiation. However, there is some overlap of magnetic susceptibility magnitudes in the four sub-units, due to their very similar petrographic nature. Sub-units 1 and 2 display a lateral zoning due to an evolution from high to medium susceptibilities, that seems to witness a magmatic diﬀerentiation from tonalitic to granodiortic compositions. Sub-unit 4 display only high susceptibilities; it is mainly made of tonalites and represents the lastemplaced magmas. The degree of anisotropy (P) ranges from 1·05 to 1·64 (Fig. 7b), with an average of 1·25. This is rather high for granitoids that are almost free of solid-state deformation microstructures, despite the fact that magnetite-bearing granites are known to be generally more anisotropic that magnetite-free granites (Bouchez 2000). The lowest values characterise the granitic core of sub-unit 3. High values characterise the eastern contact between the pluton and its country rocks. In addition, high anisotropies are also observed inside the pluton and correspond to the previously-inferred contact of sub-unit 3 against sub-unit 2. These highly anisotropic zones possibly record the forcible intrusion of magma of sub-unit 3 with respect to the more crystallised unit 2 inside the pluton, as well as some

P=K1/K3

L=K1/K2

F=K2/K3

1·27 1·50 1·35 1·23 1·14 1·16 1·45 1·09 1·40 1·29 1·19 1·29 1·11 1·39 1·38 1·49 1·55 1·35 1·25 1·18 1·31 1·14 1·27 1·09 1·29 1·20 1·27 1·46 1·19 1·35 1·26 1·11 1·28

1·06 1·29 1·09 1·06 1·04 1·09 1·21 1·04 1·19 1·08 1·04 1·04 1·07 1·26 1·16 1·18 1·15 1·06 1·09 1·06 1·10 1·05 1·08 1·05 1·16 1·02 1·07 1·10 1·13 1·17 1·05 1·05 1·10

1·21 1·21 1·26 1·17 1·10 1·07 1·24 1·03 1·21 1·21 1·15 1·25 1·04 1·13 1·22 1·31 1·40 1·29 1·16 1·12 1·21 1·09 1·19 1·04 1·13 1·17 1·18 1·32 1·06 1·18 1·21 1·06 1·14

ballooning eﬀect of the ﬁrst emplaced magma against its wall rocks. Linear and planar anisotropies are both rather high, but with a predominance of the planar anisotropies (1·14 in average for F, against 1·10 for L) suggesting the predominance of ﬂattening with respect to constrictional strain. Magnetic lineations are most valuable as evidence of mineral lineations that are generally diﬃcult to determine precisely in plutonic rocks. In the Carion pluton, magnetic lineation plunges are rather steep; this is conﬁrmed by local ﬁeld observations of the mineral lineation. Each sub-unit contains at least one zone where lineation plunges are higher than 60( (Fig. 7d). Such zones are generally regarded as the indication of magma-feeding zones (Amice & Bouchez 1989; Naba et al. 2004).

3. Interpretation 3.1. Four-pulse emplacement of the Carion magmas The Carion pluton displays a complex susceptibility zoning, that cannot ﬁt the simple normal zoning derived from preliminary geochemical studies. Moreover, the foliation pattern suggests that the pluton is made of four sub-units consistent with the distribution of susceptibility values. These sub-units are regarded as four successive magmatic pulses emplaced in a

FOUR-STAGE BUILDING OF THE CAMBRIAN CARION PLUTON

Figure 5 Projection diagrams (lower hemisphere) for all Carion pluton and country rock sampling sites. Open symbols=AMS results for each specimen; solid symbols=averages for each site.

141

142

M. O. M. RAZANATSEHENO ET AL.

recently admitted even in the absence of conspicuous internal contacts (Glazner et al. 2004).

3.2. Chronology of magma emplacement

Figure 6 Thermomagnetic curves (K vs. temperature) of representative samples from the Carion pluton.

nested fashion. Each sub-unit contains at least one zone where highly plunging lineations likely indicate the location of a magma feeder zone. Nested granites are not uncommon, but sometimes correspond to unrelated magmas with independent emplacement kinematics (Bouchez & Diot 1990). In the Carion pluton, the four magmatic pulses represent very similar magmas of intermediate composition that underwent some variable magmatic diﬀerentiation (either at depth or at emplacement level). Only in the case of sub-unit 3 did magmatic diﬀerentiation reach a proper granitic composition. Nevertheless, the whole pluton is made of co-genetic magmas as deduced from petrographic observations, very similar mineral compositions (Table 1) and whole-rock geochemistry (Fig. 4a). The magmas were likely tapped from the same source or from the same magma chamber at depth. The Carion pluton is diﬀerent from some other cases where a complex distribution pattern of susceptibility magnitudes corresponds to bimodal intrusions with mixing/mingling interactions (e.g. De´le´ris et al. 1996; Asrat et al. 2003). Indeed, the recurrence of petrogenetic processes leading to the incremental building of large plutons has been

The crosscutting relationships between foliation patterns of the four sub-units enable their emplacement order to be determined. Nevertheless, the time lag between two successive sub-units remains diﬃcult to estimate. Due to the warm environment evidenced by amphibolite to granulite facies conditions in the country rocks, the older sub-units may have remained highly ductile and even only partly crystallised until the emplacement of the younger sub-units. However, the time lag between emplacements of sub-units 2 and 3 may have been longer than in the other cases, because these sub-units are separated by a highly anisotropic area (Fig. 7b). The relative chronology of the four sub-units is consistent with the palaeomagnetic data of Meert et al. (2001b). The Carion pluton provided a good-quality palaeomagnetic pole dated at 509 (11) Ma. It was demonstrated that the remanence was a TRM carried by nearly pure magnetite, because of oven unblocking temperatures in the 560–600(C range. The 3-axis IRM experiments conﬁrmed the lack of haematite. In addition, the palaeomagnetic samples exhibit either a dual polarity, or a single reverse polarity, with an apparent spatial bias (Fig. 8). Whereas sites from the country rocks and sites from plutonic sub-units 1 and 2 display mixed polarities with reversal sequences from normal to reverse, the younger parts of the pluton, thus the latest cooled areas, i.e. the granitic core of sub-unit 3 and the sub-unit 4, only display the reverse component. Curie runs on samples that carried the reverse, normal and mixed directions showed nearly reversible heating and cooling curves with well-deﬁned Curie temperature in the magnetite range, pointing to the same type of remanence in all samples; that is, mainly a TRM carried by magnetite. Assuming parallel and rather slow cooling paths for all sub-units (a realistic assumption in such a metamorphic environment), the lack of dual polarity in the youngest parts of the plutons is evidence that these areas were slightly warmer than the others, hence too warm to register the change from ‘normal’ (negative) to ‘reverse’ (positive) inclination of the magnetic ﬁeld. Nevertheless, the lack of any signiﬁcant diﬀerence in normal and reverse polarity directions suggests that the diﬀerent plutonic sub-units (and their respective cooling paths) were separated only by relatively short time intervals. Indeed, the motion of Gondwana during the 530–490 Ma interval was quite rapid (Meert et al. 2001b) and a 5–10 Myr resolution would have been possible. Therefore, it is concluded that the whole Carion pluton was built in less than ﬁve Myrs.

3.3. Pluton-related strain vs. tectonic strain At the time of Carion pluton emplacement (5325 Ma), transcurrent tectonics along the Angavo shear zone was in its waning stage or already ﬁnished, as the Carion pluton is nearly devoid of any solid state deformation and its structures did not come into parallelism with the foliations and lineations of the shear zone. Moreover, a strain aureole with a maximum thickness of 3 kilometres can be traced in many places around the pluton. This aureole is responsible for a conspicuous foliation triple point to the south (Fig. 1c). Formation of such a triple point requires that the wall rocks were still ductile at the time of pluton emplacement (Guglielmo 1993). Indeed, new Ar–Ar ages from the Angavo shear zones point to temperatures well above 500(C at that time (Gre´goire et al. 2008). In addition, a higher degree of anisotropy is observed in the country rocks along the western border of the pluton (Fig. 7b) and is also regarded as the result of pluton-related strain.

FOUR-STAGE BUILDING OF THE CAMBRIAN CARION PLUTON

143

Figure 7 (a) Magnetic susceptibility map and location of the four sub-units. (b) Magnetic anisotropy degree map: values are presented as (P-1) 100 for sake of clarity. (c) Magnetic foliation map and projection diagram of foliation poles (lower hemisphere; contour lines at 4% and 8% for the pluton and at 2%, 4% and 8% for the country rocks). (d) Magnetic lineation map and projection diagram (lower hemisphere: contour lines at 4% and 8% for the pluton and at 2%, 4% and 8% for the country rocks).

The strain aureole is not observed along the whole northern border of the pluton. There, foliations in the wall rocks come into parallelism only with those of the last sub-unit, in agreement with its last-emplaced character. In the nearby Angavo shear zone (further East from the pluton), steeply dipping foliations are striking to the north and lineations are also trending to the north with subhorizontal to gentle plunges

(Ne´de´lec et al. 2000). Close to the Carion pluton, foliations display variable dips and lineations are more scattered (Fig. 7c–d). These observations are consistent with pluton-related strain around a post-tectonic intrusion. The pluton-related strain aureole can have resulted from a ballooning eﬀect. However, the rather steep lineations observed in the pluton also suggest that the diﬀerent magmatic

144

M. O. M. RAZANATSEHENO ET AL.

mineralisations’. J. Meert received support from a NSF grant EAR98–05306. J. L. Bouchez, K. Benn and an anonymous reviewer contributed to the substantial improvement of the manuscript. C. Cavare´-Hester, P. Lespinasse, F. de Parceval and A. M. Roquet are warmly acknowledged for technical assistance.

6. References

Figure 8 Location of the mixed and single polarity sites of Meert et al. (2001b) with respect to the four sub-units deﬁned in the present paper.

pulses may have been emplaced as individual diapirs. Diapiric emplacement cannot be advocated for granitic magmas emplaced in the brittle upper crust (Petford 1996). Here, due to the sustained high temperature conditions in the country rocks, diapiric emplacement was possible. The ﬁrst diapirically-emplaced sub-unit would have favoured the emplacement of the following magma batches, because of additional heat advection and country rock softening due to magma ascent and emplacement. Then, in agreement with the model of sequential diapirism of Weinberg (1997), the subsequent magma diapir rises into and across the former one, ﬁnally building a nested intrusion.

4. Conclusions The Carion pluton, east of Antananarivo, was emplaced as four nested magmatic sub-units identiﬁed by their magnetic foliation patterns, the complex zoning of magnetic susceptibility magnitudes and the location of the highest magnetic anisotropies along some internal contacts. Their relative chronology of emplacement is consistent with previous palaeomagnetic data showing that crystallisation of the pluton covered a time interval suﬃcient for at least one magnetic reversal, resulting in a dual polarity (from normal to reverse) in the oldest parts of the pluton and a single reverse polarity in the youngest parts of the pluton. Steep lineations and lack of solid-state deformation point to magma ascent and emplacement occurring in the waning stage of the transcurrent tectonics along the Angavo shear zone. A pluton-related strain aureole is possibly due to the diapiric ascent of the sequential magmas batches.

5. Acknowledgements The Madagascan authors received ﬁnancial support from the French Ministry of Cooperation through the Campus project ‘Geology of the Precambrian basement of Madagascar and its

Abdelsalam, M. G. & Stern, R. J. 1996. Sutures and shear zones in the Arabian-Nubian shield. Journal of African Earth Sciences 23, 289–310. Amice, M. & Bouchez, J. L. 1989. Susceptibilite´ magne´tique et zonation du batholithe granitique de Cabeza de Araya (Extremadura, Espagne). Comptes Rendus de l’Acade´mie des Sciences de Paris 308, 1171–8. Archanjo, C., Launeau, P. & Bouchez, J. L. 1995. Magnetic fabric vs. magnetite and biotite shape fabrics of the magnetite-bearing granite pluton of Gameileras (Northeast Brazil). Physics of the Earth and Planetary Interiors 89, 63–75. Asrat, A., Gleizes, G., Barbey, P. & Ayalew, D. 2003. Magma emplacement and maﬁc-felsic magma hybridization: structural evidence from the Pan-African Negash pluton, northern Ethiopia. Journal of Structural Geology, 25, 1451–69. Besairie, H. 1969. Carte ge´ologique de Madagascar a` 1/500000: feuille no 5: Tananarive. Anatananarivo, Madagascar: Service Ge´ologique. Borradaile, G. J. & Henry, B. 1997. Tectonic applications of magnetic susceptibility and its anisotropy. Earth-Science Reviews 42, 49–93. Bouchez, J. L. 1997. Granite is never isotropic: an introduction to AMS studies in granitic rocks. In Bouchez, J. L., Hutton, D. H. W. & Stephens, W. E (eds) Granite: from emplacement of melt to segregation fabric, 95–112. Dordrecht: Kluwer Academic Publishers. Bouchez, J. L. 2000. Anisotropie de susceptibilite´ magne´tique et fabrique des granites. Comptes Rendus de l’Acade´mie des Sciences de Paris 330, 1–14. Bouchez, J. L. & Diot H. 1990. Nested granites in question: contrasted emplacement kinematics of independent magmas in the Zae¨r pluton, Morocco. Geology 18, 966–9. De´le´ris, J., Ne´de´lec, A., Ferre´, E., Gleizes, G., Me´not, R. P., Obasi, C. K. & Bouchez, J. L. 1996. The Panafrican Toro Complex (northern Nigeria): magmatic interactions and structures in a bimodal intrusion. Geological Magazine 133, 535–52. Diot, H., Bolle, O., Lambert, J. M., Launeau, P. & Duchesne, J. C. 2003. The Telnes ilmenite deposit (Rogaland, South Norway): magnetic and petrofabric evidence for a Ti-enriched noritic crystal mush in a fracture zone. Journal of Structural Geology 25, 481–501. Ellwood, B. B. & Whitney, J. A. 1980. Magnetic fabrics of the Elberton granite, NE Georgia. Journal of Geophysical Research 85, 1481–6. Glazner, A. F., Bartley, J. M., Coleman, D. S., Gray, W. & Taylor, R. Z. 2004. Are plutons assembled over millions of years by amalgamation from small magma chambers. GSA Today 14, 4–11. Gleizes, G., Ne´de´lec, A., Bouchez, J. L., Autran, A. & Rochette, P. 1993. Magnetic susceptibility of the Mont-Louis Andorra ilmenite-type granite (Pyrenees): a new tool for the petrographic characterisation and regional mapping of zoned granite plutons. Journal of Geophysical Research 98, 4317–31. Gre´goire, V., Darrozes, J., Gaillot, P., Ne´de´lec, A. & Launeau, P. 1998. Magnetite grain shape fabric and distribution anisotropy vs. rocks magnetic fabric: a three-dimensional case study. Journal of Structural Geology 20, 937–44. Gre´goire, V., Monie´, P., Ne´de´lec, A. & Bouchez, J. L. 2008. Structure and thermochronology of central Madagascar: a consequence of deformation and heat advection along the late-Panafrican Angavo shear zone. Re´union des Sciences de la Terre, Nancy, April 2008. Guglielmo, G. Jr. 1993. Magmatic strains and foliation triple points of the Merrimac plutons, northern Sierra Nevada, California: implications for pluton emplacement and timing of subduction. Journal of Structural Research 15, 177–89. Hottin, A. M. 1975. Le Massif de Carion et son environnement a` l’Est de Tananarive. The`se de Doctorat de 3e`me cycle, Universite´ Paris VI, France. Hottin, G. 1976. Pre´sentation et essai d’interpre´tation du Pre´cambrien de Madagascar. Bulletin du Bureau de Recherches Ge´ologiques et Minie`res IV, 117–53.

FOUR-STAGE BUILDING OF THE CAMBRIAN CARION PLUTON Kro¨ner, A., Windley, B. F., Jaeckel, P., Brewer, T. S. & Razakamanana, T. 1999. Precambrian granites, gneisses and granulites from Madagascar: new zircon ages and regional signiﬁcance for the evolution of the Pan-African orogen. Journal of the Geological Society, London 156, 1125–35. La Roche, H (de), Leterrier, J. Grande Claude, P. & Marchal, M. 1980. A classiﬁcation of volcanic and plutonic rocks using R1–R2 diagrams and major element analysis: relationships and current nomenclature. Chemical Geology 29, 183–210. Lautel, R. 1953. Etude ge´ologique du socle cristallin a` la latitude de Tamatave. Doctoral Thesis, University of Clermont-Ferrand, France. Lawver, L. A., Gahagan, L. M. & Dalziel, I. W. D. 1998. A tight ﬁt-early Mesozoic Gondwana: a plate reconstruction perspective. Memoir of the National Institute of Polar Research, Special Issue 53, 214–29. Leake, B. E., Wooley, A. R., Birch, W. D., Burke, E. A. J., Ferraris, G., Grice, J. D., Hawthorne, F. C., Kisch, H. J., Krivovichev, V. G., Schumacher, J. C., Stephenson, N. C. N. & Whittaker, E. J. W. 1997. Nomenclature of the amphiboles: Report of the Subcomittee on Amphiboles of the International Mineralogical Association, Commission on New Minerals and Mineral Names. Canadian Mineralogist 35, 219–46. Martelat, J. E., Nicollet, C., Lardeaux, J. M., Vidal, G. & Rakotondrazafy, R. 1997. Lithospheric tectonic structures developed under high-grade metamorphism in the Southern part of Madagascar. Geodinamica Acta 10, 94–114. Meert, J. G. 2003. A synopsis of events related to the assembly of eastern Gondwana. Tectonophysics 362, 1–40. Meert, J. G., Hall, C., Ne´de´lec, A. & Madison Razanatseheno, M. O. 2001a. Cooling of a late-synorogenic pluton: evidence from laser K-feldspar modelling of the Carion granite, Madagascar. Gondwana Research 4, 541–50. Meert, J. G., Ne´de´lec, A., Hall, C., Wingate, M. & Rakotondrazafy, M. 2001b. Paleomagnetism, geochronology and tectonic implications of the Carion granite, central Madagascar. Tectonophysics 340, 1–21. Naba, S., Lompo, M., Debat, P., Bouchez, J. L. & Be´ziat, D. 2004. Structure and emplacement model for late-orogenic Paleoprotero-

145

zoic granitoids: the Tenkodogo–Yamba elongate pluton (Eastern Burkina Faso). Journal of African Earth Sciences 38, 41–57. Nagata T. 1961. Rock magnetism. Tokyo: Maruzen. 350 pp. Ne´de´lec, A., Ralison, B., Bouchez, J. L. & Gre´goire, G. 2000. Structure and metamorphism of the granitic basement around Anatananarivo: a key to the Pan-African history of central Madagascar. Tectonics 19, 997–1020. Petford, N. 1996. Dykes and diapirs? Transactions of the Royal Society of Edinburgh: Earth Sciences 87, 105–14. Rakotondrazafy, M., Madison Razanatseheno, M. O., Ne´de´lec, A., Ralison, B., Fitzsimons, I., Wingate, M. & Meert, J. G. 2001. The Cambrian Carion granite of Madagascar: a case of late PanAfrican shoshonitic magmatism. Gondwana Research 4, 746–7. Ralison, B. & Ne´de´lec, A. 1997. Contrasted Pan-African structures near Antananrivo (Madagascar). Gondwana Research Group Miscellaneous Publications (Osaka) 5, 83–4. Reeves, C. V. & de Wit, M. J. 2001. Making ends meet in Gondwana: retracing the transforms of the Indian Ocean and reconnecting continental shear zones. Terra Nova 12, 272–80. Rochette, P., Jackson, M. & Aubourg, C. 1992. Rock magnetism and the interpretation of anisotropy of magnetic susceptibility. Review of Geophysics 30, 209–26. Schmidt, M. W. 1992. Amphibole classiﬁcation in tonalite as a function of pressure: an experimental calibration of the Al-inhornblende barometer. Contributions to Mineralogy and Petrology 110, 304–10. Stern, R. J. 1994. Arc assembly and continental collision in the Neoproterozoic East African Orogen: implications for the consolidation of Gondwana. Annual Review of Earth and Planetary Sciences 22, 319–51. Thompson, A. B. & Connolly, J. A. D. 1995. Melting of the continental crust: some thermal and petrological constraints on anatexis in continental collision zones and other tectonic settings. Journal of Geophysical Research 100, 15565–79. Weinberg, R. 1997. Diapir-driven crustal convection: decompression melting, renewal of the magma source and the origin of nested plutons. Tectonophysics 271, 217–29.

MS received 22 October 2007. Accepted for publication 12 April 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 147–158, 2010 (for 2009)

Anisotropy of magnetic susceptibility fabrics in syntectonic plutons as tectonic strain markers: the example of the Canso pluton, Meguma Terrane, Nova Scotia Keith Benn Ottawa-Carleton Geoscience Centre and Department of Earth Sciences, University of Ottawa, Ottawa, ON K1N6N5, Canada Email: [email protected] ABSTRACT: The anisotropy of magnetic susceptibility (AMS) is widely and routinely used to measure the preferred orientations of Fe-rich minerals in undeformed and weakly deformed granite plutons. The interpretation of the mapped AMS fabrics depends on rock-textural observations, on the map patterns of the fabrics in plutons, and on comparisons of the pluton fabrics to tectonic structures in the country rocks. The AMS may document emplacement-ﬂow related fabrics, but the emplacement fabrics may be reworked or completely overprinted by rather weak tectonic strains of the magma mush or the cooling pluton, especially in syntectonic intrusions. The Late Devonian Canso granite pluton is an excellent example of overprinting of emplacement fabrics by weak tectonic strains. The Canso pluton was emplaced ca. 370 Ma along the boundary between the Meguma and Avalon tectonic terranes, in the northern Appalachian orogen. The AMS was mapped along two traverses that cross the pluton and that are perpendicular to the terrane boundary. Textural evidence suggests the rocks underwent very modest post-full crystallisation strains. The AMS records the dextral transcurrent shearing that occurred on the terrane boundary during emplacement and cooling of the Canso pluton, supporting interpretations that weakly deformed syntectonic granites can be used as indicators of regional bulk kinematics. AMS fabrics in Late Devonian granites of the Meguma Terrane suggest partitioning of the non-coaxial shearing into the terrane bounding fault, with pure-shear dominated deformation further from the fault. Numerical simulations suggest that the kinematics recorded by the fabrics in the Canso pluton was simpleshear, or transpression or transpression with small components of pure shear oriented perpendicular to the bounding shear zone. KEY WORDS:

AMS, Appalachians, granite, strain partitioning, syntectonic, transpression

It is common practice amongst structuralists and granitologists to classify granitic plutons as pre-, syn- and post-tectonic, referring to emplacement prior to, during or following the cessation of orogenic deformation and metamorphism of the country rocks. Such a classiﬁcation is useful in the practice of putting chronological brackets on tectonic events by isotopic dating of pluton emplacement. Interpretations of pre-, synand post-tectonic intrusions are also important in developing pluton emplacement models. Structural interpretations of pre, syn- and post-tectonic emplacement are based on observations of rock textures, the fabric pattern mapped within plutons and comparisons of pluton fabric patterns with the regional-scale structures that record tectonic events. This present paper presents new data from a syntectonic granite pluton that show tectonic overprinting of magmatic fabrics by small strains related to shearing on a nearby transcurrent terrane boundary. The Canso pluton is a granite of Late Devonian age that was emplaced near the boundary of the Meguma and Avalon terranes, in the Northern Appalachian orogen. The goals of the study are to demonstrate that the fabrics in weakly deformed plutons can provide excellent kinematic indicators of orogenic deformations, and to make the point that the fabrics must be considered in the tectonic context before they can be potentially interpreted to be

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016028

markers of magma emplacement ﬂow. Numerical simulations are also presented in order to constrain the kinematics of the tectonic overprint of magmatic fabrics in the Canso pluton. First, interpretations of fabrics in granite plutons are discussed. Then the tectonic setting of the Canso pluton is explained, the new fabric data are presented and the tectonic overprint of the igneous fabrics is argued. The numerical simulations are presented and used to constrain the kinematics along the Meguma–Avalon terrane boundary.

1. Interpreting pluton fabrics Mineral fabrics in granite plutons can be interpreted in a kinematic framework related to magma dynamics or to regional tectonics (Paterson et al. 1998). In the ﬁrst case, mineral alignments record magma dynamics driven by body forces related to transport and emplacement of the magma, or to convection driven by thermal gradients within the mostly liquid intrusion. The kinematic history recorded by the fabrics can then be interpreted to record buoyancy-driven magma ﬂow (Cruden & Launeau 1994; Tobisch & Cruden 1995) and possibly the mechanics of intrusion. In the second case, the fabric records post-emplacement tectonic deformation of the crystallising magma mush, or of the solidiﬁed pluton as it

148

KEITH BENN

Figure 1 Examples of relationships between fabrics in granite plutons and structures in the country rocks: (A) pluton fabrics are unrelated to country rock structure; (B) probable emplacement-related fabric is folded by regional deformation and lineation is reoriented parallel to the fold hinge; (C) fabric in the pluton is completely overprinted by the regional deformation; (D) pluton emplaced within a region that is undergoing bulk shearing, resulting in partial overprint of the fabric.

cooled from the solidus. The granite fabric then provides information on some (probably small) increment of the regional tectonic strain, and may also be used to infer the tectonic stress ﬁeld (Benn et al. 2001). These two cases are considered to be end-member models. In many published examples of detailed pluton-fabric studies, there is evidence of a partial tectonic overprint of a ﬁrst generation of mineral alignments that was formed during emplacement ﬂow (Bouchez & Gleizes 1995; Pignotta & Benn 1999; Z { a´k et al. 2005; C { ecˇys & Benn 2007). Figure 1 depicts four examples of fabric patterns in plutons and the relationships of the fabrics to tectonic structures in their host rocks. The precise fabric orientations that are represented in the diagrams are not of primary signiﬁcance to this discussion. The focus is on the geometric relationships of the fabrics in diﬀerent nested parts of plutons and on the relationships of pluton fabrics to the tectonic structures. The diagrams in Figure 1 are used to discuss ﬁrst order interpretations of pluton fabrics. Figure 1A–C depicts plutons emplaced into folded, or folding, country rocks, as indicated by the

anticlinal axial surface trace in each ﬁgure. Figure 1D shows another example where emplacement occurred during bulk shearing of the host rocks. Cases A and C are similar to pluton fabric development ‘completely decoupled’ from host rock deformation, and pluton fabric development ‘completely coupled’ to host rock deformation, as illustrated in ﬁgure 13A and C of Paterson et al. (1998). Figure 1 includes the necessary analyses of the mineral lineations in plutons which, in practice, are usually determined using the anisotropy of magnetic susceptibility. The interpretation of pluton fabrics is incomplete and risks being erroneous if only the mineral foliations are considered. In Figure 1A, the pluton fabrics show no geometric relationship to the regional-scale fold. There are also distinct fabric orientations in the nested portions of the pluton that would represent separate magma pulses, indicating the fabrics record the ﬂow kinematics in the two pulses. The structural relationships in Figure 1A suggest that the pluton fabrics record magma ﬂow that was decoupled from deformation in the host rocks, so they can be interpreted to indicate emplacement of the successive magma batches (Vigneresse & Bouchez 1997; Dehls et al. 1998; Siegesmund & Becker 2000; Becker et al. 2000; Molyneux & Hutton 2000; D’Eramo et al. 2006; Razanatseheno et al. 2009). This is an example of fabric in a post-tectonic pluton that can be used to infer an emplacement mechanism. Figure 1 illustrates examples where the fabrics in the plutons are reworked to varying degrees by tectonic deformation. Figure 1B shows a pluton where the foliation is rotated about the regional fold hinge, and the mineral lineation is aligned parallel to the regional hinge. This type of fabric pattern is interpreted to represent folding of a pre-existing foliation in the pluton by regional deformation and reorientation of the lineation parallel to the fold hinge. Figure 1C illustrates a foliation in the pluton that is axial-planar to the regional fold and a lineation that is parallel to the fold hinge. That example represents a stronger tectonic overprint that has resulted in parallelism of the foliation and the lineation in the pluton with the regional structures. Published examples of structural relationships as shown in Figure 1B and C have shown them to indicate emplacement during the regional folding event (Benn et al. 1997, 1998; Bolle et al. 2003; C { ecˇys & Benn 2007; Pignotta & Benn 1999). The fabrics in the plutons cannot be used to directly infer the kinematics of emplacement ﬂow, or mechanisms involved in pluton emplacement. In Figure 1D the pluton is emplaced within shearing country rocks. There is an early-formed fabric in the central parts of the pluton that may record magma emplacement ﬂow and it is variably overprinted by the regional shear. The tectonic overprint may begin as the magma in the pluton reaches some threshold of crystallisation, when the solid fraction forms a contiguous framework with increasing strength that transmits anisotropic stresses from the deforming country rocks. Examples like Figure 1D have also been shown to be indicative of syntectonic emplacement (Leblanc et al. 1996; Djouadi et al. 1997; Gleizes et al. 1998; Czeck et al. 2006; Gebelin et al. 2006). In examples such as Figure 1B and D, there may be regions of a pluton where igneous textures are preserved; but in no case should the fabrics be interpreted to indicate emplacement ﬂow kinematics, because the fabrics have likely been reworked to some degree by tectonic deformation. The following study serves to illustrate and strengthen the interpretation that, even in granites that would generally be referred to as ‘undeformed’ or ‘weakly deformed’, the fabrics may serve as tectonic strain markers.

MAGNETIC SUSCEPTIBILITY FABRICS IN THE CANSO PLUTON

149

Figure 2 Geological map of the study area, modiﬁed from Hill (1991): TSZ, Tittle shear zone; HISZ, Harbour Island shear zone, DI, Durrell’s Island. Inset: location map of the Meguma Terrane and the study area: CCFZ, Cobequid–Chedabucto fault zone; SMB, South Mountain batholith; BPP, Barrington Passage pluton.

2. Tectonic setting of the Canso pluton The Meguma Terrane of southern Nova Scotia (Fig. 2 inset) represents the most internal allochthon recognised within the northern Appalachians (Keppie 1993). The supracrustal rocks are predominantly greenschist to lower amphibolite-grade Cambro-Ordovician metaturbidites of the Meguma Supergroup, which are thought to represent detritus eroded from the Saharan Shield and deposited on the continental slope of northwestern Africa (Schenk 1971, 1995). Collision of the Meguma Terrane with the eastern margin of ancestral North America, represented by the Avalon Terrane (Fig. 2 inset), occurred in the Middle Devonian (Williams & Hatcher 1982; Keppie & Dallmeyer 1987). In Nova Scotia, the boundary between the Meguma and Avalon terranes is represented by the Cobequid–Chedabucto fault zone (CCFZ), which may be linked to a shallowly southward-dipping sole thrust underlying

the Meguma Terrane (Keppie 1993). Detailed studies along the CCFZ have revealed a prolonged history of ductile and brittle deformation related to dextral transcurrent tectonics that continued into the Carboniferous (Mawer & White 1987). The regional structural pattern within the Meguma Supergroup is dominated by upright, shallowly-plunging regional chevron folds (Horne & Culshaw 2001), and associated foliations that trend NNE–SSW in the southern part of the terrane, and that are ENE–WSW trending close to the CCFZ (Keppie 1982). These structures developed during the dextral transcurrent tectonics associated with the Meguma–Avalon collision which is recognised in Nova Scotia as the Acadian Orogeny. Acadian fabric development commenced at about 415 Ma, corresponding to the oldest 40Ar/39Ar dates obtained from whole rocks and separates of fabric-forming minerals (Dallmeyer & Keppie 1987; Keppie & Dallmeyer 1987; Muecke et al. 1988). 40Ar/39Ar dates as young as 360 Ma

150

KEITH BENN

Table 1 Characteristics of textural types. Textural type

Visible fabric

1

None

2

Weak, vertical foliation deﬁned by quartz aggregates

3

Vertical foliation and horizontal extension lineation, both deﬁned by quartz aggregates

obtained from the Meguma Supergroup (Keppie & Dallmeyer 1987) probably represent thermal resetting during the emplacement of huge volumes of peraluminous granite magma throughout the Meguma Terrane, including the South Mountain batholith (Clarke et al. 1993), the Barrington Passage pluton (Reynolds et al. 1987) and the Canso pluton, which is the subject of the present study. Geochronological data indicate that the granitic plutons of the Meguma Terrane crystallised between 380 Ma and 365 Ma (Clarke & Halliday 1980; Reynolds et al. 1981; Hill 1988; Keppie et al. 1993; Tate et al. 1997). Extensive structural, AMS and gravity modelling studies of the large, composite South Mountain batholith showed it to be composed of tabular intrusions, emplaced within a compressional tectonic stress ﬁeld, resulting in folding of emplacementrelated horizontal foliations, and lineations that are parallel to the regional fold axis (Benn et al. 1997, 1999). A detailed fabric and AMS study of the Barrington Passage pluton, in the southwestern Meguma Terrane, revealed a similar fabric pattern to the one depicted in Figure 1B (Pignotta & Benn 1999).

3. The Canso pluton and its host rocks The study area is located along the southern margin of the CCFZ, near the town of Canso, at the eastern tip of the Nova Scotian mainland (Fig. 2). The country rocks are composed of amphibolite-grade psammites and pelites of the Goldenville and Halifax Groups of the Meguma Supergroup. The rocks were strongly deformed during the Acadian Orogeny, resulting in the formation of several generations of vertical, NE–SW to ENE–WSW-trending foliations associated with tight, upright, shallowly plunging folds (Mawer & Williams 1986; Mawer & White 1987). The Canso pluton is discordant to those regional structures, but the contact-parallel folds mapped by Hill (1991) along its western margin (Fig. 2) may record shortening in the wall rocks due to expansion of the pluton during emplacement, as suggested for pluton margin parallel folds near the contact with the South Mountain batholith (Benn et al. 1999). Porphyroblasts of biotite, garnet and andalusite include dextrally crenulated segments of the regional foliation within the metamorphic aureole around the pluton, demonstrating emplacement after the development of the regional foliations (Mawer & Williams 1986; Mawer & White 1987; Hill 1991). Two vertical, E–W to WNW–ESE-striking mylonitic shear zones aﬀect the pluton (Fig. 2). Both contain horizontal extension lineations and consistently dextral shear sense indicators (C-S fabrics), showing that the granites were aﬀected by the regional dextral shearing that continued following full crystallisation, and as the pluton cooled. The Canso pluton was mapped in detail by Hill (1991), who deﬁned the lithological units shown in Figure 2. The granites tend to be homogeneous in both composition and grain size at the outcrop scale. Comagmatic enclaves are rare, and

Microstructure + Igneous texture preserved + Undulose extinction and subgrains in quartz + Quartz entirely recrystallised into roughly equant new grains, 0·2 to 0·4 mm in size, with highly lobate grain boundaries + Other minerals essentially undeformed + Quartz entirely recrystallised into elongate new grains, %0·2 mm in size, with straight or gently curved grain boundaries + Mica and feldspars deformed and locally recrystallised

centimetre-scale feldspar phenocrysts are only locally present; therefore the orientations of those markers cannot be mapped systematically. The horizontal two-dimensional nature of most outcrops (except locally along the Atlantic ocean shoreline) also makes ﬁeld mapping of mineral-preferred orientations diﬃcult. However, shallowly dipping foliations deﬁned by tabular feldspar crystals were reported for a few outcrops of undeformed granites (Hill 1988). Where autoliths of one magmatic unit are found within another, they tend to be rounded, and the strike of the pre-full crystallisation feldspar foliation is parallel in the autoliths and in the host granite, suggesting that both host and autoliths were deformed together in the magmatic state. This suggests that the diﬀerent magmatic units in Figure 2 were emplaced contemporaneously, which is consistent with similar monazite crystallisation dates obtained from the diﬀerent units of 373 Ma to 367 Ma (207Pb/235U; T. Krogh in Hill 1991). 40Ar/ 39 Ar cooling dates obtained for a relatively undeformed sample of granite vary from 366 Ma to 362 Ma (biotite and muscovite separates (Keppie & Dallmeyer 1987)), indicating that the pluton had cooled to about 300(C within %11 Ma following monazite crystallisation.

4. Rock textures Microstructures show whether the rock fabric was formed in response to deformation of the crystallising magma, or if some degree of post-full crystallisation overprint may have reworked the magmatic fabric (Paterson et al. 1989; Bouchez et al. 1990; Vernon 2000). Three microtextural types are deﬁned here for the Canso pluton. The textural types can be correlated with the strength of the mineral fabrics determined in the ﬁeld, and also using the AMS, and their distinguishing characteristics are given in Table 1. Rocks of textural type 1 are found at 14 sites, all at distances greater than 1500 m from the mylonitic shear zones. In the ﬁeld, the rock has no visible fabric, and in thin section there is evidence of only very limited post-full crystallisation strain. The igneous texture is well preserved, and grain-scale ductile strain features are limited to undulose extinction and minor subgrain development in quartz (Fig. 3A). Closer to the shear zones, there is a transition in the characteristic microstructures, recording an increase in postfull crystallisation deformation. In textural type 2, quartz is recrystallised, though micas and feldspars show little evidence of intracrystalline strain. Quartz new grains are relatively large (Table 1) and have highly lobate grain boundaries (Fig. 3B), suggestive of grain boundary migration, a high temperature recovery mechanism. The new grains form slightly elongate aggregates that deﬁne a weak, vertical foliation as observed in the ﬁeld. Closer to the shear zones, in rocks of textural type 3, the quartz aggregates are more elongate (deﬁning an easily

MAGNETIC SUSCEPTIBILITY FABRICS IN THE CANSO PLUTON

151

deformation which occurred after full crystallisation of the magma. The evidence for increasing strain and dynamic recrystallisation under decreasing temperature conditions as the shear zones are approached suggests that fabrics in rocks of textural types 2 and 3 record deformation during subsolidus cooling of the granites, and a progressive localisation of subsolidus strain into the shear zones.

5. Anisotropy of magnetic susceptibility (AMS)

Figure 3 Photomicrographs of the three microstructural types deﬁned in the text: (A) Type 1, from site 19; (B) Type 2, from site 8; (C) Type 3, from site 11.

recognised foliation in the ﬁeld) and new grains tend to be smaller than in type 2 (compare Fig. 3B and C). The quartz– quartz grain boundaries are lobate, suggesting recrystallisation under fairly high temperature conditions, but the smaller size of quartz new-grains compared to type 2 suggests greater ﬂow stresses (Urai et al. 1986) that can be attributed to minor cooling of the pluton during shearing. Some grain-scale shear instabilities (shear bands) are present in type 3 samples which is thought to be indicative of shearing in granites that have cooled to z500(C (Gapais 1989). The above observations indicate that in rocks of textural type 1, which show no sign of signiﬁcant subsolidus deformation, the preferred orientations of early formed minerals, such as biotite and euhedral feldspars, will record deformation of the granite magma. On the other hand, the clear imprint of strain and recrystallisation in textural types 2 and 3 shows that fabrics in these rocks have been reworked to some degree, by

The anisotropy of magnetic susceptibility (AMS) is a method that has proven to be eﬃcient and accurate for measuring the preferred orientations of Fe-minerals in granites, in order to map the fabrics at the pluton-scale (Bouchez 1997, 2000). Sampling for the AMS study was carried out at 30 outcrops along two traverses, mostly following roads and along the Atlantic coastline. No sampling was done for AMS within the mylonitic shear zones, as the fabrics are easily measured in the ﬁeld, and because the presence of pervasive composite planar fabrics (S-foliation and shear bands) would lead to complex magnetic fabrics (Housen et al. 1993; Aranguren et al. 1996) that are of no interest in the present study. Traverse 1 extends from 100 m to 2500 m south of the Harbour Island shear zone (HISZ, Fig. 2), with a maximum spacing between sampling sites of 300 m. Traverse 2 extends from 150 m south of the Tittle shear zone (TSZ, Fig. 2) to 1500 m north of the HISZ, with distances of 125 m to 3000 m between sites. Whereas traverse 1 is entirely within a single intrusive unit of homogeneous muscovite–biotite granite, traverse 2 crosses several internal contacts, and sampling sites therefore represent diﬀerent intrusive units. Two specimens of 22 mm-length were cut from each of three 25 mm-diameter drill cores collected per sampling site. Thus individual sites are represented by approximately 65 cm3 of rock. Individual drill cores were separated by several metres in the outcrops in order to obtain a representative sampling. The AMS was measured at the University of Ottawa using a Kappabridge KLY-2 a.c. bridge (manufactured by AGICO, Czech Republic), in a ﬁeld strength of 300 A/m at a frequency of 920 Hz. The AMS of each sample is represented by a second order tensor from which the magnitudes of the principal susceptibilities (K1RK2RK3), and the orientations of the principal directions (Kn) were obtained. Calculation of siteaverage AMS tensors and of conﬁdence ellipses about each of the average principal directions was performed using the Hext–Jelinek tensor averaging method (Jelinek 1978, 1981; Borradaile 2003). The conﬁdence ellipses represent the statistical estimates of the areas on the sphere within which 95% of the most probable average orientations of the AMS axes are included.

5.1. Source of the AMS signal The AMS of a rock sample represents the sum of the intrinsic susceptibility anisotropies of all of the constituent crystals (Hrouda 1982; Borradaile 1988; Rochette et al. 1992). The essential minerals of the Canso pluton are biotite+muscovite+ plagioclase+K-feldspar+quartz, and accessory minerals include ilmenite, apatite, monazite, garnet and zircon. The mineral assemblage is typical of the ilmenite-series of granites, where titanomagnetite is generally absent (Ishihara 1981; Barbarin 1990). Biotite makes up from 3% to 10% of the rock, and it is the most abundant mineral containing signiﬁcant amounts of Fe. Its contribution should therefore be predominant in the susceptibility signal of most samples. The susceptibility of ilmenite is greater than that of biotite (Kilmenitez 1 SI, Kbiotitez10 3 SI; Borradaile 1988), and should this

152

KEITH BENN

5.2. Results

Figure 4 Plot of the degree of magnetic anisotropy (P#%) against the susceptibility (K) for all samples. The dashed lines indicate K values for rocks where the susceptibility is entirely due to 2%, 5%, and 10% biotite, assuming susceptibility K=120010 6 SI for biotite.

mineral be locally present in more than trace amounts, it could signiﬁcantly aﬀect the AMS. We must also consider the presence of very ﬁne grained secondary oxides which may form in association with small amounts of chlorite replacing micas (Benn et al. 1993; Borradaile & Werner 1994). Figure 4 is a plot of the average susceptibility (K) for all samples against P#%=(P#1)*100. P# is the degree of anisotropy parameter suggested by Jelinek (1981), P#=exps2[1 m)2 +(2 m)2 +(3 m)2], where i =ln(Ki) (i=1 to 3), and m =3s1 ·2 ·3 . P#% was calculated following the addition of an isotropic diamagnetic (negative) susceptibility component to the measured AMS. The addition of this component is intended to eliminate the eﬀects of the diamagnetic quartz–feldspar matrix in rocks with weak magnetic susceptibility (Rochette 1987; Bouchez et al. 1990). The isotropic component Kdia = 610 6 SI corresponds to equal contributions from quartz and feldspar, based on susceptibility data reported in Borradaile (1987). Signiﬁcant contributions from locally abundant ilmenite, or secondary magnetite, should result in large variations in K in Figure 4. The susceptibilities in the Canso granites are consistently low, with 178 of 180 samples having K<10 4 SI, corresponding to expected values for rocks where K is entirely due to the presence of <10% biotite (Fig. 4). Although the magnetic susceptibility of the specimens can be mostly accounted for by the quantities of biotite present in the rocks, this may not be true for the anisotropy of the susceptibility, which can be strongly aﬀected by the presence of even small amounts of secondary magnetite (Rochette et al. 1992, 1999; Borradaile & Henry 1997). Such cases can be recognised by locally anomalous fabric orientations. To a good approximation, the AMS in the Canso granites can be interpreted in terms of the biotite petrofabric. Since K3 for biotite monocrystals is close to the [001] crystallographic axis, the measured magnetic foliation (t K3) will be parallel to the average orientation of biotite basal planes, which deﬁnes a structural foliation. As K1yK2 for biotite monocrystals, the measured magnetic lineation (K1) is parallel to the mineral lineation, which is deﬁned by the axis about which the platy biotite crystals are statistically aligned (Bouchez 1997).

The AMS data, including susceptibilities (K), the degree of anisotropy (P#%), the shape parameter for the average AMS ellipsoids (T, deﬁned in the legend of Table 2), and the orientations of the principal directions for each sampling site, are compiled in Table 2. Figure 5 presents the AMS orientation data for traverse 1. The strike of the vertical quartz–aggregate foliation, a marker of the weak subsolidus strains that aﬀected some of the sampled granites, is also shown in the orientation diagrams for those outcrops where it is developed (textural types 2 and 3, sites 6–12). Rocks of textural type 1 crop out at the nine sites that are furthest from the HISZ. Site 5 is unique in having an anomalous fabric orientation that is interpreted to indicate a strong magnetic signal from secondary oxides, and it will not be further discussed. At all other sites of textural type 1 (nos. 1–4, 25–28) the orientations of the average principal directions are very consistent. K1av (the magnetic lineation) is horizontal or shallowly plunging, with a NNE–SSW (dominantly) to NE–SW trend, which is consistently oblique in a counterclockwise sense by 50( to 75( to the HISZ. At most of the sites, the K3 principal direction is well grouped, and K3av plunges steeply, indicating a subhorizontal magnetic foliation. Approaching the shear zone, in rocks of textural types 2 and 3 the AMS fabrics are rotated in a clockwise sense, and the magnetic foliation steepens progressively. The progressive change in orientations of the fabrics is attributed to the small subsolidus strains that were undergone by the granite at those sites, as indicated by the presence of the vertical quartz– aggregate foliation. It is stressed that the quartz foliation is vertical at all of the sites, but the magnetic (biotite) foliation steepens progressively as the shear zone is approached, and as the solid-state deformation is increased, as shown by the microtextural observations. This is interpreted to indicate that the horizontal foliation that is found in the underformed rocks of textural type 1 was reworked by the solid state deformation associated with dextral shearing, and that the horizontal foliation is completely overprinted with increased strain closer to the shear zone. On traverse 2 (Fig. 6), the AMS fabric orientations deﬁne a pattern similar to the one described for traverse 1. At the sites where the rocks are of textural type 1, K3av is subvertical, indicating horizontal to shallowly-dipping magnetic foliations. K1av is subhorizontal, and trends consistently NE–SW to ENE–WSW. The orientations of K1av are counterclockwise oblique to the HISZ and the TSZ, and it rotates in a clockwise manner in the rocks of textural types 2 and 3 as the TSZ is approached. The magnetic foliation is horizontal (vertical K3av) at sites 16, 17 and 18, where the vertical quartz foliation is present. The biotite foliation, as indicated by the AMS, is perpendicular to the vertical quartz aggregate foliation that represents the superimposed subsolidus strain. This is interpreted to indicate that the subsolidus strain that formed the quartz aggregate foliation was superimposed on a horizontal biotite foliation that likely formed during magmatic deformation. The intensity of the subsolidus strain was apparently insuﬃcient to reorient the biotite foliation at those sites.

5.3. Discussion Maps of the AMS fabric orientations along two traverses through the Canso granite pluton show that an earlier-formed horizontal biotite fabric in the Canso pluton was reworked and locally overprinted by rather minor amounts of ductile strain as it cooled from the solidus. The ductile strain is certainly related to the well documented dextral shearing along the transcurrent Meguma–Avalon tectonic terrane boundary, the

MAGNETIC SUSCEPTIBILITY FABRICS IN THE CANSO PLUTON Table 2

Site average magnetic fabric data. Traverse 1

Site, Type 1, 1 2, 1 3, 1 4, 1 5, 1 6, 2 7, 2 8, 2 9, 2 10, 2 11, 3 12, 3 13, 2 14, 2 15, 2

153

K 10 6 SI 34·28 42·02 34·16 61·13 40·95 41·51 37·85 37·18 31·69 37·63 36·77 40·01 55·58 46·42 61·79

Orientations K1 K3 199/18 210/23 019/12 046/00 323/56 218/10 221/11 225/01 231/05 243/02 066/05 071/03 088/08 091/02 263/08

101/25 315/31 132/63 296/89 094/24 343/73 320/39 317/45 323/28 333/01 157/06 161/01 352/37 000/14 166/43

Traverse 2

Anisotropy P#%

Shape T

2·1 1·7 5·2 3·7 2·7 5·3 5·7 5·9 5·6 7·1 9·4 10·2 5·8 7·4 7·4

0·055 0·826 0·481 0·037 0·401 0·565 0·456 0·288 0·636 0·052 0·261 0·071 0·169 0·176 0·025

Site, Type 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30,

2 2 2 1 1 2 1 1 1 1 1 1 1 3 2

K 10 6 55·52 62·08 50·55 25·10 33·58 30·09 107·1 79·72 84·12 34·53 41·66 45·51 36·43 48·82 52·21

Orientations K1 K3 262/02 259/04 272/12 231/09 057/04 262/06 230/13 241/04 264/09 195/04 254/13 208/07 217/05 092/13 074/01

159/81 008/79 058/75 102/76 310/76 170/12 124/48 351/79 056/80 293/66 035/74 034/83 315/54 353/34 164/02

Anisotropy P#%

Shape T

9·1 7·9 11·1 2·4 4·8 3·4 6·7 4·7 4·1 6·4 3·2 2·0 5·4 7·6 6·7

0·088 0·259 0·043 0·331 0·598 0·734 0·285 0·240 0·082 0·440 0·818 0·250 0·119 0·771 0·190

Site=sampling site; Type=Textural type (see Table 1); K=magnetic susceptibility; K1, K3 =principal directions of the anisotropy of magnetic susceptibility; P#%=degree of anisotropy *100 (see text); T=2(1nK2 1nK3/(1nK1 1nK3)1 describes the shape of the AMS ellipsoid (Jelinek 1981), where T=1 for a perfectly oblate ellipsoid, and T= 1 for a perfectly prolate ellipsoid.

CCFZ. The dextral-shear overprint of the fabrics in the Canso pluton conﬁrms that dextral displacements occurred on the CCFZ at 370 Ma. The results also support the use of fabrics in weakly deformed granites to determine the bulk kinematics in orogenic belts, as previously proposed for leucogranites of the Pyrene´es mountain belt (Leblanc et al. 1996; Gleizes et al. 1997). Previous magnetic fabric studies of the South Mountain batholith (Benn et al. 1997) and of the Barrington Passage pluton (Pignotta & Benn 1999), both of which crop out within the Meguma Terrane but far from the transcurrent Meguma– Avalon terrane boundary, showed that the magnetic lineations in the plutons were reworked into orientations parallel to the regional fold axes. In the Canso pluton, which crops out quite close to the transcurrent terrane boundary, the lineations are not parallel to the regional fold hinges; instead, they are oblique to the terrane-bounding shear zone. It is proposed that this diﬀerence in the fabric record can be attributed to strain and displacement partitioning in a bulk transpressional orogen (Tikoﬀ & Teyssier 1994). In the South Mountain batholith and the Barrington Passage plutons, the fold-parallel fabrics are indicative of bulk pure shear in the folded rocks far from the terrane bounding shear zone, whereas the fabrics of the Canso pluton record the partitioning of orogen-parallel displacements and bulk non-coaxial strains into the CCFZ. Another point can be made that is of interest for interpretations of AMS fabrics in granites, in general. The fabrics in the rocks of textural type 2 have certainly been reworked by a weak tectonic strain, resulting in the reorientation of the lineations and a steepening of the foliation. In those rocks the evidence of solid-state strain is limited to the elongation of recrystallised quartz aggregates, whereas feldspar and mica crystals are for the most part undeformed. Therefore, it should be noted that the fabrics in such weakly deformed rocks cannot be easily interpreted as recording only magmatic deformation or emplacement-related strain. For example, on traverse 1, the AMS measurements of samples from several sites of textural type 1 reveal horizontal foliations and magnetic lineations that are oriented NNE– SSW, making angles as great as 75( with respect to the HISZ.

Those fabrics might be interpreted as indicating a horizontal NNE–SSW ﬂow during emplacement of the granitic magma, with the ﬂow trajectory oriented perpendicular to a feeder zone represented by the nearby CCFZ, in a manner similar to the interpretation of magnetic anisotropy fabrics in the Archean Lebel syenite pluton, Canada (Cruden & Launeau 1994). However, given the tectonic overprint of fabrics elsewhere in the Canso pluton, it is stressed that interpretations of the AMS in terms of magma emplacement, in this study, would be highly equivocal.

6. Fabric overprinting in transpression This section investigates the possibility of constraining the kinematics of the non-coaxial strain that was partitioned into the transurrent terrane boundary during advanced stages of collision of the Meguma and Avalon terranes. A recent study of syntectonic granites in the Vosges Mountains of France proved successful in modelling AMS fabric development and strain partitioning in an overall transtensional regime (Kratinova et al. 2007). A slightly diﬀerent approach to the numerical simulations is taken to the one used by those authors, but the aim of the modelling is similar. Numerical models can been applied to predict the evolutions of petrofabrics of populations of mineral grains during progressive strain histories, and the modelled petrofabrics can then be used to easily calculate the expected AMS, which can be compared to the measured AMS of rocks to infer strains and strain paths (Richter 1992; Hrouda 1993). The approach used here is similar to that used by Benn 1994, but the progressive strain is transpressional, rather than simple shear and pure shear. The preferred orientation of the c-axes of biotite crystals was measured for a sample from site CN3 (traverse 1), using a universal stage. Site CN3 is classiﬁed as textural type 1, and the AMS indicated a horizontal biotite foliation and a NNE–SSW lineation (Fig. 5). The measured biotite fabric was subjected to successive small strain increments, using the incremental strain matrix,

154

KEITH BENN

Figure 5 Map of traverse 1, south of the Harbour Island shear zone (HISZ). The number of each sampling site is indicated next to the site. The principal directions of the AMS are shown in lower-hemisphere equal-area projections. Dashed lines in some projections indicate the trace of the vertical quartz–aggregate foliation measured in outcrop. The map area is located in Figure 1.

3

⫺

0

0

1

0

0

0

⫺1

4

to simulate progressive, homogeneous strain of the biotite c-axis petrofabric. This is the kinematic model of transpression proposed by Sanderson & Marchini (1984). The dextral simple-shear component of strain takes place along vertical east–west planes, parallel to the CCFZ. represents the incremental shear strain. 1 represents the incremental stretch of a line oriented north–south horizontal, and represents the incremental stretch of a vertical line. The parameter Ti =i(1ai 1) 1 is used to indicate the relative contributions of incremental pure shear and incremental simple shear (Sanderson & Marchini 1984). Positive values

of Ti indicate ‘transpression’, in the sense that the pure shear component of incremental strain has a vertical X-axis and a horizontal, N–S oriented Z-axis. Negative values of Ti indicate ‘transtension’, where the pure shear component of incremental strain has the Z-axis vertical and the X-axis N–S horizontal. /Ti//0 indicates an increasing component of incremental pure shear and Ti =N indicates simple shear. The incremental strains were calculated with held constant at 0·01, and incremental a1 =1(/Ti). At each increment in the strain history, the AMS is calculated by summing the susceptibility tensor for each biotite crystal in the petrofabric population, assuming a perfectly oblate AMS for each crystal and an anistropy value (P#) of 1·35 (Hrouda 1993). One limitation of the approach is that the shapes of biotite crystals are not taken into account; rotations of crystals are governed by the application of the transpressional strain tensor to the vectors deﬁning the orientations of the c-axes of the

MAGNETIC SUSCEPTIBILITY FABRICS IN THE CANSO PLUTON

155

Figure 6 Map of traverse 2, south of the Tittle shear zone (TSZ). The number of each sampling site is indicated next to the site. The principal directions of the AMS are shown in lower-hemisphere equal-area projections. Dashed lines in some projections indicate the trace of the vertical quartz–aggregate foliation measured in outcrop. The map area is located in Figure 1.

crystals. In other words, the biotite crystals are assumed to behave as planes that are reoriented in response to the imposed strain (March 1931). That simpliﬁcation is not unreasonable for biotite platelets with high aspect ratios. Only small strains are considered here, so that the possibility of periodic petrofabric evolutions which may arise at high strains for populations of crystals with lower aspect ratios (Ildefonse et al. 1997) does not need to be taken into account. The modelling is intended to simulate the variations in the biotite fabric orientations and intensities documented using the AMS, from the undeformed granites of textural type 1 far from the shear zones, through textural types 2 and 3 as the shear zones are approached. An underlying assumption is that the changes in the biotite fabrics along a traverse from type 1 through type 3 represents a time- and strain-progressive overprint of the magmatic fabric at sampling site 3. In support of that interpretation it is noted that strain increases with decreasing distance from the shear zones, as indicated by the intensity of the visible fabric. Also, the microstructures suggest that deformation and dynamic recrystallisation proceeded at slightly lower temperatures as the pluton cooled. Lower-hemisphere projections of several steps in the progressive evolutions of the strain ellipsoids (modelled extension lineations and foliation poles, Fig. 7A–F) and of the K1 and K3 principal directions of the AMS (the magnetic lineation and the pole to the magnetic foliation, Fig. 7G–L) are presented for diﬀerent strain histories (transpression, transtension and

simple shear). The plots show results for steps labelled 1–4, corresponding to strain intensity values of 0·25, 0·50, 1·00, and 1·50. Strain intensities are calculated as (Nadai 1963): s =s3/3[(1 2)2 +(2 3)2 +(3 1)2]0·5 For large degrees of incremental N–S shortening, perpendicular to the terrane boundary (transpression, Ti =3, Fig. 7D), there may be a transient stage (steps 2 to 3) where the magnetic fabric path is similar to the pattern mapped in the granites, but at higher strains the magnetic lineation migrates to steeper plunges (steps 3 to 4). The modelled stretching lineation would also be vertical at relatively low strains, as previously shown by Sanderson & Marchini (1984). For large incremental stretches perpendicular to the Minas fault zone (transtension, Ti = 3, Fig. 7F), the magnetic foliation would not become vertical and such a strain path could not be responsible for the fabric pattern in the Canso granites. Progressive simple-shear (Ti =N, Fig. 7A) and progressive transpressional or transtensional strain histories with small to moderate degrees of incremental pure-shear (5%Ti%10, Fig. 7B, C; Ti = 6, Fig. 7E) all produce evolutions in the orientations of the strain axes and of the magnetic fabric axes which resemble those mapped across the strain gradients from textural type 1 to textural type 3. In each of these examples the magnetic lineation is oriented close to the stretching lineation at s%0·45. The magnetic foliation becomes vertical at strain

156

KEITH BENN

Figure 7 Results of numerical simulations of strain and fabric development during transpression and transtension. The progressive strain histories were applied to an initial biotite petrofabric that was measured for an undeformed granite sample from sample site CN3 (Fig. 5). Equal-area projections. Filled symbols are on the lower hemisphere and open symbols are on the upper hemisphere. Data are plotted for four stages of each strain history. In the magnetic fabrics diagrams, stage 0 is the AMS calculated for the initial petrofabric from sample site CN3.

intensities that are lower during progressive transpressional or simple-shear strain histories than during progressive transtensional strain histories. This is seen by comparing the data for steps 2, Ti =N, 10 and 5, with that for step 3, Ti = 6 (Fig. 7A, B, C and E; Table 2). Increasing progressive transtensional strain could result in verticalisation of the magnetic foliation and in coaxial magnetic and ﬁnite strain axes, as seen in more strongly deformed rocks near to the shear-zones (Fig. 7E). The results of the models presented here suggest that the fabric pattern in the Canso granites could be explained by a progressive strain history represented by simple-shear, or by transpression or transtension with small to moderate components of incremental shortening or stretching perpendicular to the Minas fault zone. The model results indicate that a transtensional strain history would be most likely to preserve

the early-formed horizontal magnetic foliation mapped in the rocks of textural type 2, where the magnetic lineation would have rotated to lie in the plane of the mesoscopic quartzaggregate foliation observed in the ﬁeld.

7. Conclusions The anisotropy of magnetic susceptibility (AMS) was used to map the preferred orientations of biotite crystals in a granite pluton that was emplaced during the Late Devonian Acadian Orogeny, in Nova Scotia. The pluton was emplaced near the transcurrent boundary between the Meguma and Avalon tectonic terranes. Careful consideration of the magnetic fabrics in a framework including structural, microstructural and regional tectonic information shows that the AMS fabrics

MAGNETIC SUSCEPTIBILITY FABRICS IN THE CANSO PLUTON

(inferred to indicate biotite fabrics) preserve a record of syn- to post-emplacement dextral shear on the terrane boundary. The post-emplacement shearing occurred as the pluton cooled from the solidus. An early-formed horizontal fabric, preserved at some sampling sites, was reworked and overprinted by small amounts of tectonic deformation. The results conﬁrm the usefulness of AMS fabrics in weakly deformed syntectonic plutons as markers of the bulk kinematics of orogens. Comparison of the fabrics in the Canso plutons, documented here, with previous results from plutons of similar ages but further from the terrane boundary, within the Meguma Terrane, shows that the fabrics in granites situated within diﬀerent parts of an orogen can be used to map the partitioning of coaxial and non-coaxial strains that are predicted for transpressional orogens. Numerical simulations of the fabric evolutions in the Canso pluton suggest that the kinematics of transcurrent shearing along this part of the Meguma–Avalon terrane boundary involved simple-shear or transpression, or transpression with small components of pure shear oriented perpendicular to the bounding shear zone. A markedly transpressional or transtensional progressive strain history is apparently ruled out in this study.

8. Acknowledgements The ﬁeld work for this study was funded by a Natural Sciences and Engineering Research Council of Canada (NSERC) research grant to the author. G. S. Pignotta and S. Siegesmund provided reviews that improved the paper.

9. References Aranguren, A., Cuevas, J. & Tubı´a, J. M. 1996. Composite magnetic fabrics from S–C mylonites. Journal of Structural Geology 18, 863–9. Barbarin, B. 1990. Granitoids: main petrogenetic classiﬁcations in relation to origin and tectonic setting. Geological Journal 25, 227–38. Becker, J. K., Siegesmund, S. & Jelsma, H. A. 2000. The Chinamora batholith, Zimbabwe: structure and emplacement-related magnetic fabric. Journal of Structural Geology 22 (11–12), 1837–53. Benn, K. 1994. Overprinting of magnetic fabrics in granites by small strains: numerical modelling. Tectonophysics 233, 153–62. Benn, K., Rochette, P., Bouchez, J.-L. & Hattori, K. 1993. Magnetic susceptibility, magnetic mineralogy and magnetic fabrics in a late Archean granitoid-gneiss belt. Precambrian Research 63, 59–81. Benn, K., Horne, R. J., Kontak, D. J., Pignotta, G. & Evans, N. G. 1997. Syn-Acadian emplacement model for the South Mountain Batholith, Meguma Terrane, Nova Scotia: Magnetic fabric and structural analyses. Geological Society of America Bulletin 109, 1279–93. Benn, K., Ham, N. M., Pignotta, G. S. & Bleeker, W. 1998. Emplacement and Deformation of Granites During Transpression – Magnetic Fabrics of the Archean Sparrow Pluton, Slave Province, Canada. Journal of Structural Geology 20 (9–10), 1247–59. Benn, K., Roest, W. R., Rochette, P., Evans, N. G. & Pignotta, G. S. 1999. Geophysical and structural signatures of syntectonic batholith construction: the South Mountain Batholith, Meguma Terrane, Nova Scotia. Geophysical Journal International 136, 144–58. Benn, K., Paterson, S. R., Lund, S. P., Pignotta, G. S. & Kruse, S. 2001. Magmatic fabrics in batholiths as markers of regional strains and plate kinematics: example of the Cretaceous Mt. Stuart batholith. Physics and Chemistry of the Earth 26 (4–5), 343–54. Bolle, O., Diot, H. & Trindade, R. I. F. 2003. Magnetic fabrics in the Holum granite (Vest-Agder, southernmost Norway): implications for the late evolution of the Sveconorwegian (Grenvillian) orogen of SW Scandinavia. Precambrian Research 121 (3–4), 221–49. Borradaile, G. 1987. Anisotropy of magnetic susceptibility: rock composition versus strain. Tectonophysics 138, 327–9. Borradaile, G. J. 1988. Magnetic susceptibility, petrofabrics and strain. Tectonophysics 156, 1–20.

157

Borradaile, G. J. 2003. Statistics of Earth Science data. New York: Springer-Verlag. 351 pp. Borradaile, G. J. & Henry, B. 1997. Tectonic applications of magnetic susceptibility and its anisotropy. Earth-Science Reviews 42, 49–93. Borradaile, G. J. & Werner, T. 1994. Magnetic anisotropy of some phyllosilicates. Tectonophysics 235, 223–48. Bouchez, J.-L. 1997. Granite is never isotropic: an introduction to AMS studies of granitic rocks. In Bouchez, J.-L., Hutton, D. H. W. & Stephens, W. E. (eds) Granite: from segregation of melt to emplacement fabrics, 95–112. Dordrecht: Kluwer Academic Publishers. Bouchez, J.-L. 2000. Anisotropie de susceptibilite´ magne´tique et fabrique des granites. Comptes Rendus a` l’Acade´mie des Sciences de Paris, Sciences de la Terre et des plane`tes 330, 1–14. Bouchez, J.-L., Gleizes, G., Djouadi, T. & Rochette, P. 1990. Microstructure and magnetic susceptibility applied to emplacement kinematics of granites: the example of the Foix pluton (French Pyrenees). Tectonophysics 184, 157–71. Bouchez, J.-L. & Gleizes, G. 1995. Two-stage deformation of the Mont-Louis-Andorra granite pluton (Variscan Pyrenees) inferred from magnetic susceptibility anisotropy. Journal of the Geological Society, London 152, 669–79. C { ecˇys, A. & Benn, K. 2007. Emplacement and deformation of the ca. 1.45 Ga Karlshamn granitoid pluton, southwestern Sweden, during ENE–WSW Danopolonian shortening. International Journal of Earth Sciences 96, 397–414. Clarke, D. B., Chatterjee, A. K. & Giles, P. S. 1993. Petrochemistry, tectonic history, and Sr–Nd systematics of the Liscomb Complex, Meguma Lithotectonic Zone, Nova Scotia. Canadian Journal of Earth Sciences 30, 449–64. Clarke, D. B. & Halliday, A. N. 1980. Strontium isotope geology of the South Mountain Batholith, Nova Scotia. Geochimica et Cosmochimica Acta 44, 1045–58. Cruden, A. R. & Launeau, P. 1994. Structure, magnetic fabric and emplacement of the Archean Lebel Stock, SW Abitibi Greenstone Belt. Journal of Structural Geology 16, 677–91. Czeck, D. M., Maes, S. M., Sturm, C. L. & Fein, E. M. 2006. Assessment of the relationship between emplacement of the Algoman plutons and regional deformation in the Rainy Lake region, Ontario. Canadian Journal of Earth Sciences 43 (11), 1653–71. D’Eramo, F., Pinotti, L., Tubı´a, J. M., Vegas, N., Aranguren, A. Tejero, R. & Go´mez, D. 2006. Coalescence of lateral spreading magma ascending through dykes: a mechanism to form a granite canopy (El Hongo pluton, Sierras Pampeanas, Argentina). Journal of the Geological Society, London 163, 881–92. Dallmeyer, R. D. & Keppie, J. D. 1987. Polyphase late Paleozoic tectonothermal evolution of the southwestern Meguma Terrane, Nova Scotia: evidence from Ar40/Ar39 mineral ages. Canadian Journal of Earth Sciences 24, 1242–54. Dehls, J. F., Cruden, A. R. & Vigneresse, J. L. 1998. Fracture Control of Late Archean Pluton Emplacement in the Northern Slave Province, Canada. Journal of Structural Geology 20 (9–10), 1145– 54. Djouadi, M. T., Gleizes, G., Ferre´, E., Bouchez, J. L., Caby, R. & Lesquer, A. 1997. Oblique Magmatic Structures of Two Epizonal Granite Plutons, Hoggar, Algeria – Late-Orogenic Emplacement in a Transcurrent Orogen. Tectonophysics 279 (1–4), 351–74. Gapais, D. 1989. Shear structures within deformed granites: mechanical and thermal indicators. Geology 17, 1144–7. Gebelin, A., Martelet, G., Chen, Y., Brunel, M. & Faure, M. 2006. Structure of late Variscan Millevaches leucogranite massif in the French Massif Central: AMS and gravity modelling results. Journal of Structural Geology 28 (1), 148–69. Gleizes, G., Leblanc, D. & Bouchez, J.-L. 1997. Variscan granites of the Pyrenees revisited: their role as syntectonic markers of the orogen. Terra Nova 9, 38–41. Gleizes, G., Leblanc, D., Santana, V., Olivier, P. & Bouchez, J.-L. 1998. Sigmoidal Structures Featuring Dextral Shear During Emplacement of the Hercynian Granite Complex of CauteretsPanticosa (Pyrenees). Journal of Structural Geology 20 (9–10), 1229–45. Hill, J. D. 1988. Late Devonian peraluminous granitic plutons in the Canso area, eastern Meguma Terrane, Nova Scotia. Maritime Sediments and Atlantic Geology 24, 11–19. Hill, J. D. 1991. Petrology, tectonic setting, and economic potential of Devonian peraluminous granitoid plutons in the Canso and Forest Hills areas, eastern Meguma Terrane, Nova Scotia. Geological Survey of Canada Bulletin 383. 96 pp.

158

KEITH BENN

Horne, R. & Culshaw, N. 2001. Flexural-slip folding in the Meguma Group, Nova Scotia. Canada Journal of Structural Geology 23 (10), 1631–52. Housen, B. A., Richter, C. & van der Pluijm, B. A. 1993. Composite magnetic anisotropy fabrics: experiments, numerical models, and implications for the quantiﬁcation of rock fabrics. Tectonophysics 220, 1–12. Hrouda, F. 1982. Magnetic anisotropy of rocks and its application in geology and geophysics. Geophysical Surveys 5, 37–82. Hrouda, F. 1993. Theoretical models of magnetic anisotropy to strain relationship revisited. Physics of the Earth and Planetary Interiors 77, 237–49. Ildefonse, B., Arbaret, L. & Diot, H. 1997. Rigid particles in simple shear ﬂow: is their preferred orientation periodic or steady-state? In Bouchez, J.-L., Hutton, D. H. W. & Stephens, W. E. (eds) Granite: from segregation of melt to emplacement fabrics, 177–85. Dordrecht: Kluwer Academic Publishers. Ishihara, S. 1981. The granitoid series and mineralization. Economic Geology, 75th Anniversary Volume, 458–84. Jelinek, V. 1978. Statistical processing of anisotropy of magnetic susceptibility measured on groups of specimens. Studia Geophyzika et Geodetika 22, 50–62. Jelinek, V. 1981. Characterization of magnetic fabric of rocks. Tectonophysics 79, 563–7. Keppie, J. D. 1982. Tectonic Map of the Province of Nova Scotia. Nova Scotia, Halifax: Department of Mines and Energy. Keppie, J. D. 1993. Synthesis of Palaeozoic deformational events and terrane accretion in the Canadian Appalachians. Geologisches Rundschau 82, 381–431. Keppie, J. D., Dallmeyer, R. D., Krogh, T. E. & Aftalion, A. 1993. Dating mineralization using several isotopic methods: and example from the South Mountain Batholith, Nova Scotia, Canada. Chemical Geology 103, 251–70. Keppie, J. D. & Dallmeyer, R. D. 1987. Dating transcurrent terrane accretion: an example from the Meguma and Avalon composite terranes in the northern Appalachians. Tectonics 6, 831–47. Kratinova, Z., Schulmann, K., Edel, J. B., Jezek, J. & Schaltegger, U. 2007. Model of successive granite sheet emplacement in transtensional setting: Integrated microstructural and anisotropy of magnetic susceptibility study. Tectonics 26 (6). doi:10.29/ 2006TC002035. Leblanc, D., Gleizes, G., Roux, L. & Bouchez, J.-L. 1996. Variscan dextral transpression in the French Pyrenees: new data from the Pic des Trois-Seigneurs granodiorite and its country rocks. Tectonophysics 261, 331–45. March, A. 1931. Mathematische Theorie der Regelung nach der Korngestalt bei aﬃner Deformation. Zeitschrift fu¨r Kristallographie 81, 285–97. Mawer, C. K. & White, J. C. 1987. Sense of displacement on the Cobequid–Chedabucto fault system, Nova Scotia, Canada. Canadian Journal of Earth Sciences 24, 217–23. Mawer, C. K. & Williams, P. F. 1986. Structural study of highly deformed Meguma phyllite and granite, vicinity of White Head village, S.E. Nova Scotia. Maritime Sediments and Atlantic Geology 22, 51–64. Molyneux, S.J. & Hutton, D. H. W. 2000. Evidence for signiﬁcant granite space creation by the ballooning mechanism: the example of the Ardara pluton, Ireland. Geological Society of America Bulletin 112, 1543–58. Muecke, G. K., Elias, P. & Reynolds, P. H. 1988. Hercynian/ Alleghanian overprinting of an Acadian terrane. 40Ar/39Ar studies in the Meguma Zone, Nova Scotia, Canada. Chemical Geology 73, 153–67. Nadai, A. 1963. Theory of flow and fracture of solids, Volume 2. New York: McGraw-Hill. 705 pp. Paterson, S. R., Vernon, R. H. & Tobisch, O. T. 1989. A review of criteria for the identiﬁcation of magmatic and tectonic foliations in granitoids. Journal of Structural Geology 11, 349–63. Paterson, S. R., Fowler, T. K. Jr., Schmidt, K. L., Yoshinobu, A. S., Yuan, E. S. & Miller, R. B. 1998. Interpreting magmatic fabric patterns in plutons. Lithos 44, 53–82.

Pignotta, G. S. & Benn, K. 1999. Magnetic fabric of the Barrington Passage pluton, Meguma Terrane, Nova Scotia: a two-stage fabric history of syntectonic emplacement. Tectonophysics 307 (1–2), 75–92. Razanatseheno, M. O. M., Ne´de´lec, A., Rakontondrazafy, M., Meert, J. G. & Ralison, B. 2010. Four-stage building of the Cambrian Carion pluton (Madagascar). Earth and Environmental Science Transactions of the Royal Society of Edinburgh 100(for 2009), 133–45. Reynolds, P. H., Zentilli, M. & Muecke, G. K. 1981. K–Ar and 40 Ar/39Ar geochronology of granitoid rocks from southern Nova Scotia: Its bearing on the geological evolution of the Meguma Zone of the Appalachians. Canadian Journal of Earth Sciences 18, 386–94. Reynolds, P. H., Elias, P., Muecke, G. K. & Grist, A. M. 1987. Thermal history of the southwestern Meguma zone, Nova Scotia, from an 40Ar/39Ar and ﬁssion track dating of intrusive rocks. Canadian Journal of Earth Sciences 24, 1952–65. Richter, C. 1992. Particle motion and the modelling of strain response in magnetic fabrics. Geophysical Journal International 110, 451– 64. Rochette, P. 1987. Magnetic susceptibility of the rock matrix related to magnetic fabric studies. Journal of Structural Geology 9, 1015–20. Rochette, P., Jackson, M. & Aubourg, C. 1992. Rock magnetism and the interpretation of anisotropy of magnetic susceptibility. Reviews of Geophysics, 30, 209–26. Rochette, P., Aubourg, C. & Perrin, M. 1999. Is this magnetic fabric normal? A review and case studies in volcanic formations. Tectonophysics 307 (1–2), 219–34. Sanderson, D. J. & Marchini, W. R. D. 1984. Transpression. Journal of Structural Geology 6, 449–58. Schenk, P. E. 1971. Southeastern Atlantic Canada, northwestern Africa, and continental drift. Canadian Journal of Earth Sciences 8, 1218–35. Schenk, P. E. 1995. Meguma Zone. In Williams, H. (ed.) Geology of the Appalachian–Caledonian Orogen in Canada and Greenland. Geology of Canada, 261–77. Ottawa: Geological Survey of Canada. Siegesmund, S. & Becker, J. K. 2000. Emplacement of the Ardara pluton (Ireland): new constraints from magnetic fabrics, rock fabrics and age dating. International Journal of Earth Sciences 89, 307–27. Tate, M. C., Clarke, D. B. & Heaman, L. M. 1997. Progressive hybridisation between Late Devonian maﬁc-intermediate and felsic magmas in the Meguma Zone of Nova Scotia, Canada. Contributions to Mineralogy and Petrology 126, 401–15. Tikoﬀ, B. & Teyssier, C. 1994. Strain modeling of displacement-ﬁeld partitioning in transpressional orogens. Journal of Structural Geology 16, 1575–88. Tobisch, O. T. & Cruden, A. R. 1995. Fracture-controlled magma conduits in an obliquely convergent continental magmatic arc. Geology 23, 941–4. Urai, J. L., Means, W. D. & Lister, G. S. 1986. Dynamic recrystallization of minerals. In Hobbs, B. E. & Heard, H. C. (eds) Mineral and Rock Deformation: Laboratory Studies. Geophysical Monograph, 161–99. Washington: American Geophysical Union. Vernon, R. H. 2000. Review of microstructural evidence of magmatic and solid-state ﬂow. Electronic Geosciences 5 (2): http:// link.springer-ny.com/link/service/journals/10069/bibs/0005001/ 00050002.htm. Vigneresse, J.-L. & Bouchez, J.-L. 1997. Successive Granitic Magma Batches During Pluton Emplacement – the Case of Cabeza De Araya (Spain). Journal of Petrology, 38 (12), 1767–76. Williams, H. & Hatcher, R. H. J. 1982. Suspect terranes and accretionary history of the Appalachian orogen. Geology 10, 530–6. Z { a´k, J., Schulmann, K. & Hrouda, F. 2005. Multiple magmatic fabrics in the Sazava pluton (Bohemian Massif, Czech Republic): a result of superposition of wrench-dominated regional transpression on ﬁnal emplacement. Journal of Structural Geology 27 (5), 805–22.

MS received 29 January 20008. Accepted for publication 30 April 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 159–172, 2010 (for 2009)

Chemical structure in granitic magmas – a signal from the source? J. D. Clemens1, P. A. Helps2 and G. Stevens1 1

Department of Earth Sciences, University of Stellenbosch, Private Bag X1, 7602 Matieland, South Africa Email: [email protected]

2

School of Geography, Geology and the Environment, CEESR, Kingston University, Penrhyn Rd, Kingston-upon-Thames, Surrey KT1 2EE, UK

ABSTRACT: Though typically exhibiting considerable scatter, geochemical variations in granitic plutons and silicic volcanic deposits are commonly modelled as products of diﬀerentiation of originally homogeneous magmas. However, many silicic igneous bodies, particularly those classiﬁed as S-types, are internally heterogeneous in their mineralogy, geochemistry and isotope ratios, on scales from hundreds of metres down to one metre or less. The preservation of these heterogeneities supports recent models for the construction of granitic magma bodies through incremental additions of numerous batches (pulses) of magma derived from contrasting sources. Such pulses result from the sequential nature of the melting reactions and the commonly layered structure of crustal magma sources. Internal diﬀerentiation of these batches occurs, but not generally on the scales of whole magma chambers. Rather than being created through diﬀerentiation or hybridisation processes, at or near emplacement levels, much of the variation within such bodies (e.g. trace-element or Mg# variation with SiO2 or isotope ratios) is a primary or near-source feature. At emplacement levels, the relatively high magma viscosities and slow diﬀusion rates of many chemical components in silicic melts probably inhibit processes that would lead to homogenisation. This permits at least partial preservation of the primary heterogeneities. KEY WORDS:

chemical variation, diﬀerentiation, granitic magma, inherited heterogeneity

Granitic and silicic volcanic rocks are a major component of the Earth’s continental crust, and occur in a wide variety of tectonic settings. The geochemical variations found within granitic plutons and silicic volcanic deposits are commonly modelled as the result of processes that occur, during ascent and crystallisation, to modify the composition of an originally relatively homogeneous parent magma. Examples of such processes include various magma hybridisation phenomena (e.g. assimilation of wall rocks and magma mixing) and diﬀerentiation processes (e.g. crystal fractionation, crystal unmixing and ﬂuid-driven leaching). These processes can result in chemical variation, and the production of a series of genetically related magma fractions. They can and do dominate the production of variation within some suites of maﬁc to intermediate igneous rocks, in which very well deﬁned chemical variation trends are observed, and are supported by textural evidence for crystal-liquid separation. However, how generally realistic is the common approach of extending this set of models to more silicic magma systems? This paper deﬁnes silicic magmas as those that crystallise to ‘granitic’ rocks that would be classiﬁed (Le Maitre 1989) as alkali feldspar granites, syenogranites, monzogranites, granodiorites and trondhjemites. Rocks of this type dominate the world’s silicic batholiths and volcanic complexes, in a wide range of ages and tectonic settings. Using a variety of evidence, it will be shown that there is reason to conclude that much geochemical variation in suites of silicic igneous rocks probably has its origin in primary variations inherited from the magma source regions. For silicic magmas, diﬀerentiation, hybridisation, etc. are secondary processes, commonly acting only at relatively small scales.

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016053

1. Previous work Numerous previous studies have amply demonstrated the composite character of some silicic volcanic complexes and granitic bodies. In the volcanic context, a notable example is the work of Mills et al. (1997), Huysken et al. (2001) and Bindeman & Valley (2003) on the large-volume pyroclastic sequences of the SW Nevada Volcanic Field. These authors presented evidence for the eruption of magma batches with contrasting major-element, trace-element and isotopic characteristics, from the same set of vents, over geologically very short times, with aggregate magma volumes of the order of 1000 km3. In the plutonic context, Mohr (1991) studied the cryptic variations in Sr and Nd isotope ratios within the sheets that make up the huge, S-type Leinster batholith of Ireland, and concluded that the variations were inherited from the magma source region and that interactions between magma batches were minimal. Structural studies have identiﬁed many instances of incremental magma addition to growing plutons, leading to internally sheeted bodies with textural, mineralogical, geochemical and isotopic contrasts between adjacent sheets (e.g. McNulty et al. 1996; Paterson & Miller 1998; Wiebe & Collins 1998; Johnson et al. 1999; Miller & Paterson 2001; Weinberg et al. 2001; Mahan et al. 2003; Archanjo & Fetter 2004; Belcher & Kisters 2006). An example of this type of heterogeneity in a body apparently lacking sheeted structure is the Phillips pluton in Maine, studied by Pressley & Brown (1999). This body is composed of leucogranite enclosing centimetric to decametric masses of granodiorite. The two granitic compositions have contrasting Nd isotope ratios, and are

160

J. D. CLEMENS ET AL.

inferred to have been derived from diﬀerent rock types in a single heterogeneous source region. Pressley & Brown concluded that there had been no signiﬁcant mixing between the magmas. White et al. (2001) came to a similar conclusion in considering Sr isotope heterogeneity in the relatively maﬁc S-type Jillamatong pluton of southeastern Australia. They regarded the lack of correlation between SiO2 content and initial 87Sr/86Sr as evidence against any form of magma mixing. Scaillet et al. (1996) calculated the viscosities of Himalayan S-type leucogranites under likely temperature and ﬂuid conditions during the genesis of the magmas by partial melting of metasediments. The viscosity that they found is around 3104 Pa.s. They concluded that laminar ﬂow in feeder dikes of the size actually observed in the Himalaya would cause source-derived heterogeneities in initial 87Sr/86Sr to be preserved, even if multiple, contrasting magma batches ascended as a single physical batch in the same conduit. These examples are all from S-type granitic rocks, and perhaps these might be expected to display such heterogeneity, given the strongly chemically layered nature of many clastic sedimentary accumulations. However, examples will be provided of I-type rocks that show similar features. The question posed is whether geochemical variation within these identiﬁed, isotopically distinct magma batches could also have a primary origin, rather than represent evidence for fractionation.

2. Geological context of the present study This study draws on the authors’ own petrological and geochemical data from a variety of high-level granitic and volcanic complexes of Archaean and Palaeozoic age, emplaced in tectonic settings varying from transcurrent shearing, to compressional terrane accretion and post-orogenic extension. Both I- and S-type examples are included, that vary in composition from granites and monzogranites to trondhjemites and granodiorites. The Stolzburg pluton is a 3·45 Ga trondhjemite (Kamo & Davis 1994) that intrudes the Theespruit Formation, the stratigraphically lowermost unit in the Barberton Greenstone Belt of South Africa. Its tectonic setting is somewhat obscured by younger metamorphic events, but it has been suggested to represent a group of syntectonic sheets intruded during 3·45 Ga compressional accretion (de Wit et al. 1987; de Ronde & de Wit 1994). The Nelshoogte pluton is a synorogenic 32361 Ma trondhjemite/tonalite body (de Ronde & Kamo 2000) that intruded the SW margin of the Belt during terrane amalgamation and metamorphism (Kisters et al. 2003). Both the Stolzburg and Nelshoogte plutons are characterised by exceptionally low degrees of strain, in the context of the Greenstone Belt, though both plutons have strong gneissosity in places, predominantly conﬁned to the margins (e.g. Kisters et al. 2003). The composite Criﬀell pluton was emplaced at 400 Ma into lower Palaeozoic formations of the Scottish Southern Uplands, during post-orogenic extension. It is an essentially undeformed, high-level, I-type body that contains little textural evidence of hybridisation or contamination (e.g. magmatic enclaves or xenoliths). Previous work has revealed an apparently concentric zoning pattern, in the form of a discontinuous outer margin of metaluminous hornblende granodiorite and a core of increasingly peraluminous muscovite-biotite granite (Stephens et al. 1985). Stephens & Halliday (1980) suggested that the granodiorite and the granite represent two separate magma pulses, and Stephens (1992) concluded that compositional variations within Criﬀell represent the aggregated eﬀects of between-pulse and within-pulse variation, the between-pulse variation being generally dominant.

The 373 Ma Violet Town Volcanics of Central Victoria, in SE Australia (Clemens & Wall 1984) are undeformed, garnetand cordierite-bearing, S-type, rhyolitic to rhyodacitic, intracaldera ignimbrites, emplaced subaerially onto the eroded surface of very low-grade Silurian to mid-Devonian marine metasediments, in a post-orogenic extensional setting, associated with development of red-bed basins. The Dartmoor pluton, the largest granitic body in the Cornubian batholith of SW Britain, was emplaced into lowgrade metasediments of Carboniferous age. Like the previous two groups of rocks, it was emplaced in a setting of postorogenic extension. This granite is a peraluminous, S-type, emplaced at 280 Ma (Darbyshire & Shepherd 1985). Brammall & Harwood (1923) identiﬁed four main, textural facies that were sequentially intruded; a biotite microgranodiorite, a coarse-grained K–feldspar–phyric biotite granite, an equigranular, sparsely megacrystic granite, and various minor aplitic and microgranitic dykes. Contacts between the diﬀerent facies are sharp.

3. Petrographic and textural evidence Many granitic intrusions are internally heterogeneous, in the sense that they contain rocks with a wide range of compositions. Contacts between these rock fractions are seldom gradational over more than a few metres. Commonly, these internal contacts are quite sharp and clearly indicate intrusion of one phase by another, or several others. This sort of relationship suggests the successive intrusion of contrasting magma batches. These batches could be interpreted as magma fractions produced by diﬀerentiation elsewhere and then successively emplaced at the site of intrusion. However, there is evidence to the contrary, as will be seen below. Chappell & White (1974) and Chappell (1984) deﬁned Sand I-type granites. Whatever may be thought of that classiﬁcation, it cannot be denied that diﬀerent granitic magmas can be derived from quite contrasting source materials. Thus, it is noteworthy that there are numerous examples of granitic intrusions that are composite, in the sense that I- and S-type magma fractions clearly occupied diﬀerent parts of the shared space in the pluton. In the context of granitic complexes that are either purely Sor purely I-type, Belusova et al. (2006) studied the internal morphology, geochemistry and Hf and Pb isotope systematics of zircons from, amongst other rocks, a sample of the S-type Numbla Vale monzogranite from the Berridale batholith in southeastern Australia. The zoning within the zircons here indicates that there must have been several contrasting magma batches added to the pluton, from at least two contrasting sources. Thus, there is clear evidence that magmas with contrasting sources were involved in the construction of some granitic intrusions, and that the variations are not always readily apparent from ﬁeld and petrographic examination of the rocks. Certainly, this variability is commonly on a large scale, and it is possible that the smaller magnitude variations, within these regionally extensive magma batches, could be due to diﬀerentiation processes. Again, however, evidence will be presented that such smaller-scale variations too are commonly not produced by in situ processes. Large, intrusive, maﬁc to ultramaﬁc bodies very commonly show signs of regionally extensive layering, sometimes cyclic in character. In these occurrences, less dense minerals commonly occupy the upper parts of the individual layers and the denser phases the lower, with the implication that gravity has played a major role in producing these structures. There are examples of layering caused by repeated injection of fresh magma, but

CHEMICAL STRUCTURE IN GRANITIC MAGMAS

161

in situ diﬀerentiation seems to be the most plausible explanation for the origins of the stratiﬁcation within the diﬀerent layers. In contrast to the maﬁc and ultramaﬁc intrusions, non-tectonised silicic intrusive rocks only very rarely exhibit any form of igneous layering. When it is present, it occurs on small scales (up to a few metres) and has been interpreted as the result of minor ﬂow segregation of large and dense crystals suspended in the moving magma. Such local segregation processes cannot have been responsible for the production of large-scale variability in the compositions of the rocks within a granitic pluton. There is thus a fundamental diﬀerence in the mechanisms controlling crystal distribution within maﬁc and felsic magmas – probably due to the small but important contrast in the viscosities of the two sorts of magma (e.g. Clemens & Petford 1999), as well the signiﬁcant diﬀerence in density contrast between the major rock-forming minerals in the two types of magmas. As has often been pointed out, and recently in the analysis by Whittington & Treloar (2002), the densities and viscosities of granitic magmas are strong functions of H2O content in the melt. However, the work of Clemens & Petford (1999) and Clemens & Watkins (2001) showed the fundamental interdependence of temperature and melt H2O content in the great majority of granitoid magmas. This means that the full range of possible viscosities and densities is not realised in nature and the inﬂuences of these parameters is therefore relatively small in purely felsic magma systems.

4. Major- and trace-element evidence In maﬁc to intermediate rock series, SiO2 content is a useful proxy for the degree of fractionation, so Harker plots portray the major- and trace-element variation very well. It is well known that groups of genetically related maﬁc to intermediate rocks show a strong tendency to form tight arrays in such plots (e.g. for TiO2, Al2O3, FeO*, MgO, CaO and P2O5, and trace elements such as Ni, Cr, Sr and Zr). Some scatter is probably caused by diﬀerences in proportions of fractionating mineral phases and by analytical errors. Nevertheless, well deﬁned trends are the norm. In contrast, the degree of scatter observed in Harker plots of the same elements in many silicic rock suites is much greater, even when the ﬁeld relations suggest that the rocks are genetically related. Where it occurs, this scatter should not be dismissed as noise or analytical error. It is suggested that this feature of the geochemical data for silicic series points to magma features or processes that contrast with those in more maﬁc systems. The question arises: what should chemical variation look like in a suite of felsic rocks produced by diﬀerentiation from a single parent magma? The best way to illustrate this is to examine inter-element plots for intrusive units for which all the geochemical and isotopic evidence points to in situ diﬀerentiation as the origin of the variation. One well-documented example is the I-type Boggy Plain pluton in south-eastern Australia (Wyborn et al. 2001). Figure 1 shows F (total Fe as FeO+MgO) plotted against TiO2 and K2O in the Boggy Plain rocks. The tightness of the variation here is shown by a large number of other oxides and elements, and is typical of rock series produced through fractionation processes. It shows more similarity with the variation in maﬁc suites than it does with typical felsic suites, which exhibit a much greater degree of scatter in the data. Indeed, here it appears that the felsic magmas are the products of diﬀerentiation of a parent maﬁc magma. Another excellent example for felsic magmas in which crystal fractionation has controlled the chemical variation has

Figure 1 Selected inter-element correlation plots for the Boggy Plain pluton, redrawn from ﬁgure 3 of Wyborn et al. (2001).

recently been reported by Rodrı´guez et al. (2007). These authors describe isotopically homogeneous dacitic magmas that vary from about 62 to 66 wt% SiO2. The parent magmas were apparently mantle-derived hybrids of andesitic composition and, like the Boggy Plain pluton, the Harker plots for major- and trace-element variations exhibit tight quasi-linear trends, produced through hornblende fractionation (Fig. 2). This variation was successfully modelled for both major and trace elements, including REE. The preceding examples of tight chemical variations in felsic magmas, apparently due to diﬀerentiation by crystal fractionation, have important features in common. Such variations seem to occur in felsic systems derived by diﬀerentiation from maﬁc to intermediate parent magmas that are either mantle derived or mainly mantle-derived. In contrast, granitic magmas derived from crustal, or mainly crustal, sources typically show far less well-deﬁned variations on a variety of geochemical plots. Some examples are provided below. Figure 3 shows Harker plots for Mg#, CaO and P2O5 for the Stolzburg and Nelshoogte plutons, introduced earlier. The data deﬁne the type of rough negative trends that are typical for many crustally-derived granitic rock series. Particularly in the case of the Nelshoogte data, it might be argued that this negative correlation is a magma evolutionary trend of some kind. However, it is noteworthy that the Stolzburg and Nelshoogte data generally plot in the same area of the diagrams, and that the combined dataset deﬁnes more deﬁnite trends than does that for either individual pluton. The magma batches represented by these two plutons diﬀer in age by w200 Ma and are related only by the fact that they most probably formed through partial melting of similar, hydrated, metabasic source rocks (Clemens et al. 2006). Magmas formed

162

J. D. CLEMENS ET AL.

Figure 2 Harker plots showing some major-element variations in the volcanic products of Nevado de Longavı´ Volcano in central Chile (modiﬁed from ﬁgure 7 of Rodrı´guez et al. 2007).

from similar sources, under similar conditions will show similar geochemical variations, even if the magmas are unrelated, because the melting equilibria will produce granitic melts, of very similar composition, that coexist with very similar residual mineral assemblages. The variation within and between such bodies as the Stolzburg and Nelshoogte plutons could be controlled by processes that operated at or near the magma sources (rather than at the near-emplacement stage). This possibility is strengthened by the fact that the observed variation in trace elements hosted only by major minerals is diﬃcult to model by fractionation of any combination of the major minerals present in the rocks. In particular, the variations in Rb, Sr and Ba (in feldspars and micas) cannot be modelled satisfactorily as the results of fractionation of any viable phase or mineral assemblage (Clemens et al. 2006). Following on from the approach above, geochemical plots can be used to model the eﬀects of fractionation of the observed crystalline phases in silicic rock suites on major- and trace-element contents of the evolved magmas. This is done by calculating vectors for the fractionation of each mineral and comparing the results to the observed variations, essentially using the approach of Rollinson (1993). Many silicic suites

have been tested this way and it has been found that only rarely do predicted trends match the observed data. Usually, no combination of fractionating phases can describe the data adequately. It is emphasised that this problem lies with major elements and trace elements hosted by the major mineral phases, not with trace elements controlled by the distribution of accessory minerals. Thus, in models for the genesis and chemical evolution of granitic magmas, there is a deep systematic mismatch between the trends expected for fractional crystallisation and the actual variations recorded in rocks crystallised from crustally derived granitic magmas. Figure 4 illustrates this problem for the rocks of the Dartmoor pluton, also introduced earlier. The ﬁgure shows the eﬀects of mineral fractionation on the Sr, Rb and Ba concentrations in the residual magma. The plots show the directions and magnitudes of vectors that would be produced by the fractionation of 10% each of biotite, K-feldspar and plagioclase. It must be stressed that there are always uncertainties in the values of crystal–liquid distribution coeﬃcients for natural magma systems, leading to uncertainty in the results of any such modelling. However, the present authors have experimented with the distribution coeﬃcients, varying them within

CHEMICAL STRUCTURE IN GRANITIC MAGMAS

163

Figure 3 Harker plots for Mg#, CaO and P2O5 for the Stolzburg and Nelshoogte plutons.

the bands of quoted values (Whittington & Treloar 2002), and the conclusions reached seem relatively robust. The Ba–Sr plot (Fig. 4a) rules out biotite and plagioclase fractionation, as enormous degrees of fractionation of any combination of these phases would be required. Orthopyroxene, a near-liquidus phase in many S-type granitic magmas, could be considered as a fractionating phase instead of biotite (with which Opx is replaced during later magma evolution). However, as can be seen from the plot, Opx is even less eﬀective than biotite, producing a trend in the wrong sense. At

ﬁrst sight, Figure 4a suggests the possibility of K-feldspar fractionation. However, it is apparent that >20% K-feldspar fractionation would be required to explain the spread of rock compositions in Dartmoor, something that cannot be the case, given the petrological evidence. First, there is not enough phenocrystal K-feldspar present in the rocks to account for such a trend. Secondly, ﬁeld observations have not revealed the presence of any strong mechanical accumulations of K-feldspar. Finally, the late crystallisation of K-feldspar in S-type granitic magmas is well established from experimental

164

J. D. CLEMENS ET AL.

Figure 4 Log–log plots showing (a) Ba–Sr and (b) Rb–Sr variations in the rocks of the Dartmoor pluton of the Cornubian batholith in south-western Britain. Vectors are shown for the eﬀects of fractionation of 10% each of biotite, K-feldspar and plagioclase (minerals present in the rocks) and 10% of orthopyroxene, which might have been present at an earlier stage of magma evolution. The standard Rayleigh fractionation equation was used (C=C0·fD1), where C is the instantaneous concentration of the element, C0 is the initial concentration, f is the fraction of liquid remaining (0·9) and D is the crystal–liquid distribution coeﬃcient. Mineral–melt element distribution coeﬃcients for orthopyroxene were taken from Arth (1976). For biotite, the values (DRb =2, DSr =0·04) were taken from Icenhower & London (1995), with DBa set at 10, reﬂecting the moderate temperature probable for the Dartmoor magmas, and consistent with the Arth (1976) value. For plagioclase, the expressions of Blundy & Wood (1991) were used, with the composition set at An36 (the measured composition of the plagioclase crystal cores in the more maﬁc Dartmoor rocks) and calculated for a magma at 900(C. The resulting values are; DRb =0·09, DSr =5·8 and DBa =0·69. For K-feldspar (Or85), the expressions of Icenhower & London (1995) were used, resulting in DRb =0·88, DSr =12 and DBa =21·3.

studies (e.g. Clemens & Wall 1981), and is also apparent from the petrographic examination of the Dartmoor rocks. As in most granitic rocks, the majority of the K-feldspar present is interstitial to the earlier plagioclase, quartz and biotite. Crystallising at such a late stage, and from what amounts to an interstitial, residual liquid, it cannot have controlled any fractionation processes. Thus, K-feldspar could not have been present either suﬃciently early in the crystallisation of the magmas, or in suﬃcient quantity to be responsible for fractionation. Any combination of the mineral vectors also results in impossible quantities of fractionated phases. Figure 4b, the Rb–Sr plot, shows that a combination of plagioclase, K-feldspar and biotite fractionation could theoretically produce the trend but, again, the amounts of fractionated phases are infeasible. Note also the degree of scatter in this plot, despite the fact that the axes are logarithmic. This mismatch between trace-element variations and the mineralogical, petrological and experimental evidence strongly implies a deep, possibly source-inherited origin for much of the geochemical variation within Dartmoor. These major variations are inconsistent with crystal fractionation at any depth. Fractionation of early Opx, Grt, etc. would result in evolutionary trends completely contrary to those observed. Note also that the use of the chosen data sources for construction of Figure 4 results in relatively high distribution coeﬃcients for feldspars, favourable to explanation of the trends as fractionation. The failure of these models is thus even more telling. Variation due to mineral fractionation is very probably present in these rocks as well, but it seems to have been superimposed on and to be subordinate to the main variations inherited from processes that occurred at deeper levels, prior to magma emplacement. In an intriguing use of a geochemical normal

probability plot to study incompatible element variation in silicic ignimbrites from Costa Rica, Vogel et al. (2004, p. 155) found that ‘The normal probability distributions of Nb/Ta from . . . closely related erupted units are distinct and represent diﬀerent populations . . . Fractional crystallisation could not have produced this variation among the units . . .’. This ﬁnding is in accord with what is presented in the present paper, and it is suggested that this mismatch is a prevalent feature of crustally derived granitic magmas, although this needs to be veriﬁed for a wider variety of felsic rock units.

5. Isotope evidence Ratios of stable isotopes (e.g. 18O and 34S) and radiogenic isotopes (e.g. initial 87Sr/86Sr, 143Nd/144Nd and 206Pb/204Pb) are well established tracers of rock parentage. For many granitic bodies, true whole-rock Rb–Sr isochrons cannot be produced, and there are commonly signiﬁcant disagreements between Rb–Sr dates and more accurate and precise Pb–Pb dates based on sub-grain-scale analysis of zircon, monazite, etc. These discrepancies occur in rocks with isotope systems undisturbed by alteration and they are therefore signiﬁcant. Their existence suggests that the diﬀerent rock types in many granitic plutons have commonly not been formed by diﬀerentiation of any original, homogeneous parent magma (the assumption used in constructing whole-rock isochrons), but are more probably produced from isotopically contrasting magma batches that were aggregated together to form the pluton. In granitic plutons with overt variation from I- to S-type character, the isotope ratios are usually strongly correlated with the variations in mineral assemblages characteristic of the

CHEMICAL STRUCTURE IN GRANITIC MAGMAS

two sorts of magma/rock. This clearly indicates that such plutons were ﬁlled with contrasting batches of magma derived from diﬀerent sources. However, there is emerging evidence that signiﬁcant degrees of isotopic variation (far in excess of typical analytical error) can also be found within what appear to be relatively homogeneous masses of granitic rock, that show no mineralogical signs of multiple parentage. Some examples are provided below. It is instructive to consider the scales on which such cryptic variation can occur. Helps et al. (2003) carried out detailed sampling of a number of granitic plutons, and presented preliminary results of high-precision analyses for initial 87Sr/ 86 Sr and 143Nd/144Nd, and 18O. In the S-type Dartmoor granite, they discovered variation (an order of magnitude greater than analytical error) on length scales as small as 0·5 m, in rocks with no discernible petrographic variations. In the I-type Criﬀell pluton similar variation is present, but only at scales >100 m. Further evidence of isotopic heterogeneity in granitic magmas can be found in the study of Waight et al. (2000), who used microdrilling techniques to sample zoned plagioclase crystals in granites. These authors discovered signiﬁcant variations in initial 87Sr/86Sr between adjacent zones in the crystals. Such variation shows that these crystals grew from batches of melt that had diﬀerent isotopic compositions, either by the crystals freely circulating through contrasting magma batches, or by growing from contrasting melts that percolated through a relatively static mush of growing crystals. Such isotopic variability is a clear indication that granitic bodies were constructed by the agglomeration of multiple fractions of contrasting magma. These disparate magma fractions may have been derived from diﬀerent materials in a heterogeneous magma source, or may result from modiﬁcation of magma composition by a range of hybridisation mechanisms at some depth greater than the ﬁnal emplacement level. The implication is that what may appear to be magma fractions generated by diﬀerentiation can represent chemically and isotopically distinct magma batches that were produced at or near the magma source region. Mingling and mixing of these magma batches undoubtedly occurs, due to chaotic ﬂow in the composite magma, but the preservation of these heterogeneities highlights the ineﬃciency of mingling and diﬀusive reequilibration in felsic systems.

6. A commentary on some recent literature studies There have been numerous recent studies of granitic complexes that show isotopic and trace-element heterogeneity, apparently unrelated to physical evidence for magma mingling. Nevertheless, mingling and mixing processes are the most typical models presented for these occurrences. Cristoﬁdes et al. (2007) studied the 50 Ma, metaluminous Sithonia Plutonic Complex of northern Greece. Here there are dioritic to gabbroic masses, hornblende tonalites with maﬁc enclaves (monzonitic to dioritic and tonalitic), hornblende tonalites, hornblende and biotite granodiorites, leucogranites, two-mica granites and aplitic to pegmatitic dykes. The published model involves two-stage magma mixing with crystal fractionation – ‘The ﬁrst step explains the chemical variation in the maﬁc enclave group: a basic magma, represented by the least evolved enclaves, interacted with an acid magma, represented by the most evolved granitoid rocks, to give the most evolved enclaves. The second step explains the geochemical variations of the remaining rocks of the basic group: most evolved enclaves interacted with the same acid magma to give the spectrum of rock compositions with intermediate geochemical signatures.’ (Cristoﬁdes et al. 2007, p. 243). Figure 5 was

165

constructed using the Sr concentration and Sr isotope data presented in table 4 of this reference. Mixing lines can be readily identiﬁed using such diagrams, and it is apparent that, although magma mixing might have been involved here, the published scheme is untenable. Interaction between the least evolved enclave magmas and the most evolved granitic magmas would not produce any of the trends observed for the more maﬁc rocks. However, there may have been mixing between tonalitic host magma and an enclave magma to produce the more evolved enclaves. The second step in the model is equally untenable. The ‘acid magma’ would appear to be a range of rather contrasting felsic entities. Fractionation of a tonalitic parent could have produced the hornblende–biotite granodiorites, but production of the rest of the series would have to involve mixing with at least three diﬀerent more felsic end-members (a two-mica granite, a leucogranite and a separate aplitic magma). Aplitic magmas are commonly thought of as diﬀerentiates of granitic parents, but here they would appear to have been derived from a separate parent magma, unrelated to the other rocks in the Complex. The dioritic to gabbroic rocks plot in a part of the diagram that suggests that they are unrelated to any of the other rocks in the Complex; they are certainly not candidates for the parents of the enclaves or their host tonalites. Figure 6 shows the K2O Harker plot from ﬁgure 4 of Cristoﬁdes et al. (2007). Again, this shows that the diorites and gabbros appear to be a distinct magma, which may have fractionated to produce the range of compositions in this group, but not related to the rest of the magmas. The enclaves show trends that could be related to hybridisation with the tonalites, though this is not a simple relationship. The rest of the graph is not particularly helpful in revealing genetic relationships between the rock groups, and nor are the other Harker plots published in ﬁgure 4 of Cristoﬁdes et al. (2007). As pointed out by Clemens (1989, p. 1315) although they portray the variations very well, ‘Harker diagrams are close to impotent as tools for process diagnosis’. A signiﬁcant feature of Figure 6 is the high degree of scatter in the data, typically with 250% variation in K2O content at any given SiO2 value. Taken together, these data do not support a fractionation model for the origin of the variation. Indeed, a recent investigation into melt compositions produced from various metasedimentary sources (Stevens et al. 2007) has demonstrated that alkali metal and Ca concentrations in the melt vary substantially, reﬂecting variations in the feldspar mineral assemblage and the feldspar compositions in the sources. Deep mixing processes may have been involved in the evolution of the Sithonia magmas but, if so, there were clearly several end-members. Certainly, the primary heterogeneity here is one of deep, probably near-source origin, rather than something that developed at near-emplacement levels. Hildreth & Wilson (2007) presented a mass of data on the Bishop Tuﬀ, erupted from the Long Valley magma chamber in eastern California, at 760 ka. The Tuﬀ is zoned in composition, crystal content and temperature, but not due to any shallow diﬀerentiation processes. The pre-eruption magma chamber was evidently ﬁlled incrementally, with numerous batches of contrasting, crystal-poor rhyolitic magma. These batches were derived from a deep source (or sources), retained their geochemical character during high-level crystallisation (phenocryst formation) and were preserved as distinct pumice units after eruption. The heterogeneity in the chemistry of the pumice magmas is illustrated in Figure 7. The plot shows the variations in CaO, TiO2, Zr and Ba, all of which are nonvolatile elements whose concentrations should not be aﬀected by ﬂuid–ﬂuxed processes. The high degree of scatter is apparent, even within any given textural variety of pumice. Hildreth

166

J. D. CLEMENS ET AL.

Figure 5 Sr plotted against 1000/Sr for the rocks of the Sithonia plutonic complex (N. Greece), using data from Cristoﬁdes et al. (2007). Symbols are: +-shaped crosses=diorites and gabbros; ﬁlled circles and open circles= monzonitic and dioritic to tonalitic enclaves, respectively; ﬁlled squares=tonalites; open squares and triangles=hornblende- and biotite-granodiorites, respectively; stars=leucogranites; diamonds=two-mica granites; -shaped crosses=aplites and pegmatites.

& Wilson ascribe this scatter to fractionation processes operating in a mush zone beneath the volcanic magma chamber, within which magma batches evolved and were then delivered, crystal-poor, to a high-level volcanic magma chamber. This mush zone is envisaged to exist at mid-crustal depths. An equally valid view of this would be that the mush zone represents a heterogeneous magma source, with highly variable melt proportions and diﬀerent degrees of residual solid entrainment into the magmas withdrawn from it. In any case, here too, the primary variation is of deep origin and was not developed through crystal fractionation at or near magma emplacement level. Recently, Stevens et al. (2007) investigated the origin of major-element variations in the S-type, leucogranitic to granodioritic rocks of the Cape Granite Suite, South Africa. These authors proposed that the generally linear trends evident in Harker diagrams for this group of rocks result entirely from processes that operated in the magma source region. In the present paper, the interrelationships between major elements

are proposed to result from variable degrees of entrainment, into the magmas, of the peritectic products of the biotite melting reactions that produced the magmas. The more leucocratic granites are proposed to be close to pure melts, bearing only accessory minerals, whilst the more maﬁc rocks in the suite represent mixtures between melts and up to 20% of the peritectic products (mainly garnet and Fe–Ti oxides). Interestingly, the factors that favour peritectic phase entrainment at the source are also those considered to favour accessory phase entrainment. This would explain the commonly observed correlations between major oxides (e.g. FeO) that could reﬂect peritectic phase entrainment, and trace elements (e.g. Zr) that reﬂect accessory phase entrainment and/or dissolution. Stevens et al. (2007) make a clear distinction between this highly selective incorporation of peritectic and some accessory minerals and the bulk incorporation of solids, as in the restite unmixing model of Chappell et al. (1987). The mechanism proposed by Stevens et al. (2007) could produce magmas (and rock series) characterised by tight inter-element correlations,

CHEMICAL STRUCTURE IN GRANITIC MAGMAS

167

Figure 6 K2O Harker plot for the rocks of the Sithonia plutonic complex (N. Greece), modiﬁed from part of ﬁgure 4 of Cristoﬁdes et al. (2007). Symbols are as in Figure 5.

especially in cases where there is little variation in melt compositions, without the requirement of mechanically segregating minerals. Indeed, the low density contrasts and relatively high viscosities, that hamper fractional crystallisation in felsic magmas, would enhance the eﬃciency of peritectic phase entrainment, and permit this mechanism to shape the compositions of some granitic rock suites.

7. A magnetic perspective The evidence above suggests that many granitic plutons were constructed through the aggregation of a number of magma batches or pulses. Similar conclusions were reached by other workers, recently, for example, by Glazner et al. (2004). In some cases, these magma batches were derived either from diﬀerent source regions or from diﬀerent parts of a heterogeneous source region. Evidently, the processes of magma ﬂow within the pluton, reaction and diﬀusion are ineﬃcient, and complete homogenisation seems to be uncommon. At least in some felsic magmatic bodies, the persistence of these chemically and isotopically contrasting magma batches probably explains a good deal of the variation. However, although we can sample the rocks and detect the existence of these batches or heterogeneities, unless they have clear mineralogical or structural expressions, it is very diﬃcult or impossible to determine their physical extent and distribution using geochemistry. Using grid sampling it is possible to detect largescale variations, sometimes cryptic, (e.g. Stephens & Halliday 1980; Stephens et al. 1985; Stephens 1992; Glazner et al. 2004), but there seems to be evidence of quite small-scale, completely cryptic variation that would not be detected unless outcropscale grid sampling and analysis were carried out – an expensive and speculative proposition. The best hope for detecting complex magma heterogeneities and mapping their extent probably lies in geophysics, rather than geochemistry. The aim here would be to map gravity or magnetic variations, not to reveal magma emplacement mechanisms, but to track the extents of magma batches with contrasting properties, and possibly reveal how the magma batches ﬂowed into and round the pluton (magma chamber). Geophysical imaging from airborne surveys, particularly magnetic imaging, can provide a useful way of examining structures within rock bodies, provided that they contain

Figure 7 Harker plots for some non-volatile major and trace elements in the Bishop Tuﬀ pumices, modiﬁed from parts of ﬁgures 9 and 10 of Hildreth and Wilson (2007).

heterogeneities marked by suﬃcient contrasts in magnetic susceptibility and remanence. This is particularly likely to be the case in composite S-type (ilmenite- and pyrrhotite-bearing)

168

J. D. CLEMENS ET AL.

Figure 8 Map of a section of northwestern Victoria, Australia, showing the ﬁrst vertical derivative of total magnetic intensity. The lighter the grey, the higher the magnetic susceptibility of the rocks. The map shows some of the internal structure of Devonian granitic plutons (Dg=unknown type; Dgi=I-type; and Dgs=S-type) that intrude Cambrian ( Csm) and Ordovician (unlabelled) low-grade metasediments. This is a greyscale version of a section of the 1:1 000 000 scale map of Simons and Moore (1999), published as an enclosure in VandenBerg et al. (2000).

and some I-type (more oxidised and sometimes magnetitebearing) plutons. However, this technique has been largely neglected by igneous petrologists working on the genesis of granitic bodies. Figure 8 illustrates the kind of image that could prove rather enlightening in discussions such as that in the present work. The plutons shown in Figure 8 are Devonian I-type granitic bodies in the northwestern part of the state of Victoria, in

Australia. Unfortunately, these plutons are largely unexposed, due to a thin but ubiquitous cover of Quaternary ﬂuvial sediments from the Murray River. The magnetic image cuts through this veil of young sediments to reveal spectacular lobate and concentric ring patterns in the magnetic susceptibility of the underlying granitic rocks. These patterns are due to variations in the magnetic mineral contents of the rocks and are not at all what might be expected to result from in situ

CHEMICAL STRUCTURE IN GRANITIC MAGMAS

169

Figure 9 Diagram of the map in Figure 8, showing tentative assignments of magma sources (stars) and ﬂow patterns (arrows). These are highly speculative, but stars without arrows are meant to indicate possible magma upwelling sites at which subsequent magma ﬂow occurred somewhat more equally in all radial directions. Arrows not emanating from stars are meant to suggest the traces of earlier ﬂow patterns, the sources of which appear to have been overprinted by ﬂows from later pulses.

diﬀerentiation by gravity or side-wall accretion. The vertical derivative, used in this map of the anomalies, reﬂects the rate-of-change of anomalies (mainly with depth), as well as their strengths. It is likely to include signals from depths up to a few km, biased toward any shallower features. The banding and blotching in Figure 8 may well be the magnetic signature of the batches or pulses of magma that we have been discussing. Furthermore, the patterns here are strongly reminiscent of ﬂow features in lavas, and it is a simple exercise to identify

potential feeder zones from which the magma batches may have emanated and ﬂowed laterally to feed the growing plutons. Tentative assignments of such feeder zones and ﬂow directions are shown in Figure 9. The higher- and lowersusceptibility bands could indicate sub-pulses of magma within the tongues or lobes that may correspond to the main pulses of magma (rather like ﬂow units in lavas). Whatever the case, these features are undeniably of interest and should be further investigated.

170

J. D. CLEMENS ET AL.

8. Conclusions This survey of characteristics of silicic magmatic rocks shows that granitic plutons and silicic volcanic complexes are very commonly composite bodies, constructed by the accumulation of batches (pulses) of magma derived from diﬀerent sources. The evidence presented in the present paper suggests that this heterogeneity is independent of the oft-observed magma mingling that is due to the late incorporation of more maﬁc enclave magmas. It most probably reﬂects variations in the source rocks that were partially melted to form the various magma batches or deep-seated (near-source) mingling of magmas from diﬀerent protoliths. Derivation of magma fractions through diﬀerentiation of an initially homogeneous parent magma may be common in maﬁc to intermediate systems, but appears to be less so in silicic systems, especially those formed from magmas generated by crustal melting. This should be no surprise as, even in some maﬁc magmas, initial, source-related heterogeneities are preserved through later magma evolution (e.g. Bryce & DePaolo 2004). Diﬀerentiation in granitic magmas can lead to production of extreme compositional variants, commonly enriched in Li, Be, B, F, etc. Such diﬀerentiation most probably occurs by mechanisms such as crystal fractionation, but the kilometric- to regional-scale variations in many granitic plutons (such as those in Fig. 8) are of deep origin and most commonly primary (source-inherited). This leads to a large degree of scatter in Harker variation diagrams, with the primary heterogeneity further complicated by contrasting diﬀerentiation paths that are followed within local magma batches. The heterogeneities involved are sometimes expressed as separate intrusive sheets, dykes or blobs, zoning of plutons or layering of volcanic sequences, or they may be more occult (visible only in the isotopic compositions or with subtle mineralogical expressions). The causes of this primary heterogeneity in silicic magmas include such processes as: + partial melting of compositionally contrasting sectors of a single source region + migration (with time) of a melting front into rocks of contrasting composition + sampling of melts from diﬀerent levels, by magma ascending in a sub-vertical conduit through a partially molten region + weakening of a migrating thermal pulse, allowing only more fertile rocks to partially melt as time progresses + diﬀering degrees of magma mixing and/or wall-rock assimilation at deep levels in the crust. Whatever combination of these mechanisms may be responsible for the heterogeneity in a given case, it seems that the diﬀerent magma pulses are commonly channelled upward through the same magma plumbing system, to form a particular pluton or volcanic deposit. If we accept that a good deal of the variation among the rocks of silicic igneous complexes is formed in this way, rather than by diﬀerentiation at emplacement levels, we need to be circumspect about the use of whole-rock isochrons for dating such bodies. This is especially so if assembly of the complexes took a signiﬁcant period of time, within the resolution of the dating technique. In addition, the preservation of pre-emplacement heterogeneities implies that granitic magma chambers commonly do not experience chamber-wide, crystal–liquid processes. It seems more probable that chemical evolution of the magmas occurs within individual magma batches that can have limited volumes and areal extents. At emplacement levels, processes leading to magma homogenisation (e.g. magma mixing, convective stirring and diﬀusive re-equilibration) must be ineﬃcient; they are probably inhibited by the relatively high magma viscosities and

the slow diﬀusion rates of many chemical components in silicic melts (as discussed, for example, by Sparks & Marshall 1986). Using ﬁeld and geophysical evidence, Glazner et al. (2004) and Parada et al. (2005) also concluded that large and broadly homogeneous plutons can accumulate incrementally, from many smaller magma batches. Such pulses were predicted by Clemens & Mawer (1992), based on consideration of the mechanics of dyke ascent of granitic magma. Bons et al. (2004) extended this physical concept in presenting a simple numerical model for melt segregation and batch transport by fracture propagation in a self-organised critical state. Bons et al. (2004) also recognised the potential for geochemical complexity in the resulting pluton, if the batches were not subsequently homogenised. However, pulsed melt production (as diﬀerent hydrous mineral-bearing assemblages successively break down or as the locus of melting shifts in space and time) could also be responsible for magma pulses and heterogeneity (e.g. Clemens 2005). The construction of a pluton from multiple pulses of magma clearly has major implications for the thermal structure and cooling history of the body. Annen et al. (2006) emphasised this aspect in their recent paper on the genesis of intermediate to silicic magmas in what they term ‘deep crustal hot zones’. Annen et al. (2006) did comment (p. 533) that rejuvenation of former magma batches, by later injections, could lead to ‘extremely complicated radiometric data’. Glazner et al. (2004, p. 4) stated that ‘. . . many aspects of the petrochemical evolution of magmatic systems (e.g. in situ crystal fractionation and magma mixing) need to be reconsidered.’ The present authors would agree. There is a general need to reassess the meaning of the geochemical and isotopic data for felsic magmatic systems.

9. Acknowledgements JC’s work for this paper was enabled partly through sabbatical leave support from the Faculty of Science and the Geodynamics and Crustal Processes Research Group (within the Centre for Earth and Environmental Sciences Research) at Kingston University (London, UK). Reviews by Nick Petford, Bill Collins, Steve Sparks, Colin Donaldson, Alan Whittington and Fritz Finger were helpful, but the views expressed here are those of the authors and should not necessarily be ascribed to any of these reviewers.

10. References Annen, C., Blundy, J. D. & Sparks, R. S. J. 2006. The genesis of intermediate and silicic magmas in deep crustal hot zones. Journal of Petrology 47, 505–39. Archanjo, C. J. & Fetter, A. H. 2004. Emplacement setting of the granite sheeted pluton of Esperanc¸a Brasiliano orogen, Northeastern Brazil. Precambrian Research 135, 193–215. Arth, J. G. 1976. Behaviour of trace elements during magmatic processes. United States Geological Survey Journal of Research 4, 41–7. Belcher, R. W. & Kisters, A. 2006. Progressive adjustments of ascent and emplacement controls during incremental construction of the 3·1 Ga Heerenveen batholith, South Africa. Journal of Structural Geology 28, 1406–21. Belusova, E. A., Griﬃn, W. L. & O’Reilly, S. Y. 2006. Zircon crystal morphology, trace element signatures and Hf isotope composition as a tool for petrogenetic modelling: examples from eastern Australian granitoids. Journal of Petrology 47, 329–53. Bindeman, I. N. & Valley, J. W. 2003. Rapid generation of both highand low-delta O-18, large-volume silicic magmas at the Timber Mountain/Oasis Valley caldera complex, Nevada. Geological Society of America Bulletin 115, 581–95. Blundy, J. D. & Wood, B. J. 1991. Crystal-chemical controls on the partitioning of Sr and Ba between plagioclase feldspar, silicate melts, and hydrothermal solutions. Geochimica et Cosmochimica Acta 55, 193–209.

CHEMICAL STRUCTURE IN GRANITIC MAGMAS Bons, P. D., Arnold, J., Elburg, M. A., Kalda, J., Soesoo, A. & Van Milligen, B. P. 2004. Melt extraction and accumulation from partially molten rocks. Lithos 78, 25–42. Brammall, A. & Harwood, H. F. 1923. The Dartmoor granite: its mineralogy, structure and petrology. Mineralogical Magazine 20, 39–53. Bryce, J. G. & Depaolo, D. J. 2004. Pb isotopic heterogeneity in basaltic phenocrysts. Geochimica et Cosmochimica Acta 68, 4453– 68. Chappell, B. W. 1984. Source rocks of I- and S-type granites in the Lachlan Fold Belt, southeastern Australia. Philosophical Transactions of the Royal Society of London A310, 693–707. Chappell, B. W., White, A. J. R. & Wyborn, D. 1987. The importance of residual source material (restite) in granite petrogenesis. Journal of Petrology 28, 1111–38. Chappell, B. W. & White, A. J. R. 1974. Two contrasting granite types. Pacific Geology 8, 173–4. Clemens, J. D. 1989. The importance of residual source material restite in granite petrogenesis: a comment. Journal of Petrology 30, 1313–16. Clemens, J. D. 2005. Melting of the continental crust I: ﬂuid regimes, melting reactions and source-rock fertility. In Brown, M. & Rushmer, T. (eds) Evolution and diﬀerentiation of the continental crust, 297–331. Cambridge: Cambridge University Press. Clemens, J. D., Yearron, L. M. & Stevens, G. 2006. Barberton South Africa. TTG magmas: Geochemical and experimental constraints on source-rock petrology, pressure of formation and tectonic setting. Precambrian Research 151, 53–78. Clemens, J. D. & Mawer, C. K. 1992. Granitic magma transport by fracture propagation. Tectonophysics 204, 339–60. Clemens, J. D. & Petford, N. 1999. Granitic melt viscosity and silicic magma dynamics in contrasting tectonic settings. Journal of the Geological Society, London 156, 1057–60. Clemens, J. D. & Wall, V. J. 1981. Crystallization and origin of some peraluminous S-type. granitic magmas. Canadian Mineralogist 19, 111–31. Clemens, J. D. & Wall, V. J. 1984. Origin and evolution of a peraluminous silicic ignimbrite suite: the Violet Town Volcanics. Contributions to Mineralogy and Petrology 88, 354–71. Clemens, J. D. & Watkins, J. M. 2001. The ﬂuid regime of hightemperature metamorphism during granitoid magma genesis. Contributions to Mineralogy and Petrology 140, 600–6. Cristoﬁdes, G., Perugini, D., Koroneo, A., Soldatos, T., Poli, G., Eleftheriadis, G., Del Moro, A. & Neiva, A. M. 2007. Interplay between geochemistry and magma dynamics during magma interaction: An example from the Sithonia Plutonic Complex NE Greece. Lithos 95, 243–66. Darbyshire, D. P. F. & Shepherd, T. J. 1985. Chronology of granite magmatism and associated mineralization, SW England. Journal of the Geological Society, London 142, 1159–77. de Ronde, C. E. J. & de Wit, M. J. 1994. Tectonic history of the Barberton greenstone belt, South Africa: 490 million years of Archean crustal evolution. Tectonics 13, 983–1015. de Ronde, C. E. J. & Kamo, S. L. 2000. An Archaean arc–arc collisional event: a short-lived ca. 3 Myr. episode, Weltevreden area, Barberton greenstone belt, South Africa. Journal of African Earth Sciences 30, 219–48. de Wit, M. J., Armstrong, R. A., Hart, R. J. & Wilson, A. H. 1987. Felsic igneous rocks within the 3·3 to 3·5 Ga Barberton greenstone belt: high crustal level equivalents of the surrounding tonalite–trondhjemite terrain emplaced during thrusting. Tectonics 6, 529–49. Glazner, A. F., Bartley, J. M., Coleman, D. S., Gray, W. & Taylor, R. Z. 2004. Are plutons assembled over millions of years by amalgamation from small magma chambers? GSA Today 14, 4–11. Helps, A., Clemens, J. D. & Petford, N. 2003. Source-related chemical and isotopic heterogeneities in granitoids. Geophysical Research Abstracts 5, 02661. Hildreth, W. & Wilson, C. J. N. 2007. Compositional zoning of the Bishop Tuﬀ. Journal of Petrology 48, 951–99. Huysken, K. T., Vogel, T. A. & Layer, W. 2001. Tephra sequences as indicators of magma evolution: Ar-40/Ar-39 ages and geochemistry of tephra sequences in the southwest Nevada volcanic ﬁeld. Journal of Volcanology and Geothermal Research 106, 85–110. Icenhower, J. & London, D. 1996. Experimental partitioning of Rb, Cs, Sr, and Ba between alkali feldspar and peraluminous melt. American Mineralogist 81, 719–34. Johnson, S. E., Paterson, S. R. & Tate, M. C. 1999. Structure and emplacement history of a multiple-center, cone-sheet-bearing ring

171

complex: The Zarza Intrusive Complex, Baja California, Mexico. Geological Society of America Bulletin 111, 607–19. Kamo, S. L. & Davis, D. W. 1994. Reassessment of Archean crustal development in the Barberton Mountain Land, South Africa, based on U–Pb dating. Tectonics 13, 167–92. Kisters, A. F. M., Stevens, G., Dziggel, A. & Armstrong, R. A. 2003. Extensional detachment faulting and core-complex formation in the southern Barberton granite-greenstone terrain, South Africa: Evidence for a 3·2 Ga orogenic collapse. Precambrian Research 127, 355–78. Le Maitre, R. W. 1989. A classification of igneous rocks and glossary of terms. Oxford: Blackwell. 193 pp. Mahan, K. H., Bartley, J. M., Coleman, D. S., Glazner, A. F. & Carl, B. S. 2003. Sheeted intrusion of the synkinematic McDoogle pluton, Sierra Nevada, California. Geological Society of America Bulletin 115, 1570–82. McNulty, B. A., Tong, W. & Tobisch, O. T. 1996. Assembly of a dike-fed magma chamber: the Jackass Lakes pluton, central Sierra Nevada, California. Geological Society of America Bulletin 108, 926–40. Miller, R. B. & Paterson, S. R. 2001. Construction of mid-crustal sheeted plutons: examples from the North Cascades, Washington. Geological Society of America Bulletin 113, 1423–42. Mills, J. G., Saltoun, B. W. & Vogel, T. A. 1997. Magma batches in the timber mountain magmatic system, southwestern Nevada volcanic ﬁeld, Nevada, USA. Journal of Volcanology and Geothermal Research 78, 185–208. Mohr, P. 1991. Cryptic Sr and Nd Isotopic Variation Across the Leinster Granite, Southeast Ireland. Geological Magazine 128, 251–6. Parada, M. A., Roperch, P., Guiresse, C. & Ramirez, E. 2005. Magnetic fabrics and compositional evidence for the construction of the Caleu pluton by multiple injections, Coastal Range of central Chile. Tectonophysics 399, 399–420. Paterson, S. R. & Miller, R. B. 1998. Mid-crustal magmatic sheets in the Cascades Mountains, Washington: Implications for magma ascent. Journal of Structural Geology 20, 1345–63. Pressley, R. A. & Brown, M. 1999. The Phillips pluton, Maine, USA: evidence of heterogeneous crustal sources and implications for granite ascent and emplacement mechanisms in convergent orogens. Lithos 46, 335–66. Rodrı´guez, C., Selle´s, D., Dungan, M., Langmuir, C. & Leeman, W. 2007. Adakitic dacites formed by intracrustal crystal fractionation of water-rich parent magmas at Nevado de Longavı´ Volcano 36.2(S; Andean Southern Volcanic Zone, Central Chile. Journal of Petrology 48, 2033–61. Rollinson, H. R. 1993. Using geochemical data: evaluation, presentation, interpretation, 352 pp. London: Longman Scientiﬁc & Technical. Scaillet, B., Holtz, F., Pichavant, M. & Schmidt, M. 1996. Viscosity of Himalayan leucogranites: implications for mechanisms of granitic magma ascent. Journal of Geophysical Research B101, 27691–9. Simons, B. A. & Moore, D. H. 1999. Victoria 1:1 000 000 Pre-Permian geology. Melbourne, Australia: Geological Survey of Victoria. Sparks, R. S. J. & Marshall, L. A. 1986. Thermal and mechanical constraints on mixing between maﬁc and silicic magmas. Journal of Volcanology and Geothermal Research 29, 99–124. Stephens, W. E. & Halliday, A. N. 1980. Discontinuities in the composition surface of a zoned pluton, Criﬀell, Scotland. Geological Society of America Bulletin 91, 165–70. Stephens, W. E., Whitley, J. E., Thirlwall, M. F. & Halliday, A. N. 1985. The Criﬀell zoned pluton – correlated behavior of rareearth element abundances with isotopic systems. Contributions to Mineralogy and Petrology 89, 226–38. Stephens, W. E. 1992. Spatial, compositional and rheological constraints on the origin of zoning in the Criﬀell pluton, Scotland. Transactions of the Royal Society of Edinburgh: Earth Sciences 83, 191–9. Stevens, G., Villaros, A. & Moyen, J-F. 2007. Selective peritectic garnet entrainment as the origin of geochemical diversity in S-type granites. Geology 35, 9–12. Vandenberg, A. H. M., Willman, C. E., Maher, S., Simons, B. A., Cayley, R. A., Taylor, D. H., Morand, J., Moore, D. H. & Radojkovic, A. 2000. The Tasman Fold Belt system in Victoria. Geological Survey of Victoria Special Publication, 483 pp. Melbourne, Australia: Geological Survey of Victoria. Vogel, T. A., Patin˜o, L. C., Alvarado, G. E. & Gans, B. 2004. Silicic ignimbrites within the Costa Rican volcanic front: evidence for the formation of continental crust. Earth and Planetary Science Letters 226, 149–59.

172

J. D. CLEMENS ET AL.

Waight, T. E., Maas, R. & Nicholls, I. A. 2000. Fingerprinting feldspar phenocrysts using crystal isotopic composition stratigraphy: implications for crystal transfer and magma mingling in S-type granites. Contributions to Mineralogy and Petrology 139, 227–39. Weinberg, R. F., Sial, A. N. & Pessoa, R. R. 2001. Magma ﬂow within the Tavares pluton, northeastern Brazil: compositional and thermal convection. Geological Society of America Bulletin 113, 508–20. White, A. J. R., Allen, C. M., Beams, S. D., Carr, F., Champion, D. C., Chappell, B. W., Wyborn, D. & Wyborn, L. A. I. 2001. Granite suites and supersuites of eastern Australia. Australian Journal of Earth Sciences 48, 515–30.

Whittington, A. G. & Treloar, P. J. 2002. Crustal anatexis and its relation to the exhumation of collisional orogenic belts, with particular reference to the Himalaya. Mineralogical Magazine 66, 53–91. Wiebe, R. A. & Collins, W. J. 1998. Depositional features and stratigraphic sections in granitic plutons: implications for the emplacement and crystallization of granitic magma. Journal of Structural Geology 20, 1273–89. Wyborn, D., Chappell, B. W. & James, M. 2001. Examples of convective fractionation in high-temperature granites from the Lachlan Fold Belt. Australian Journal of Earth Sciences 48, 531–41.

MS received 28 March 2008. Accepted for publication 27 May 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 173–183, 2010 (for 2009)

Causes of compositional diversity in a lobe of the Half Dome granodiorite, Tuolumne Batholith, Central Sierra Nevada, California R. C. Economos1, V. Memeti1, S. R. Paterson1, J. S. Miller2, S. Erdmann3 and J. Z { a´k4,5 1

University of Southern California, Department of Earth Sciences, 3651 Trousdale Parkway, Los Angeles, CA 90089–0740, USA Email: [email protected]

2

Department of Geology, San Jose´ State University, San Jose´, CA 95192–0102, USA

3

Department of Earth Sciences, Dalhousie University, Halifax, Nova Scotia, B3H 4R2 Canada

4

Institute of Geology and Paleontology, Charles University, Albertov 6, Prague, 12843, Czech Republic

5

Czech Geological Survey, Kla´rov 3, Prague, 11821, Czech Republic

ABSTRACT: The causes of compositional diversity in the Tuolumne Batholith, whether source heterogeneity, magma mixing, or fractional crystallisation, is a matter of longstanding debate. This paper presents data from detailed mapping and a microstructural and major element, trace element and isotopic study of an elongate lobe of the Half Dome granodiorite that protrudes from the southern end of the batholith. The lobe is normally zoned from quartz diorite along the outer margin to high-silica leucogranite in the core. Contacts are steep and gradational, except for the central leucogranite contact, which is locally sharp: magmatic fabrics overprint contacts. A striking feature of the lobe is the 18 wt% SiO2 range comparable to that observed for the entire Tuolumne Batholith. Feldspar-compatible elements (Sr and Ba) decrease towards the centre, while Rb increases. Light and middle REEs show a smooth decrease towards the centre of the lobe. Calculated initial isotopic ratios of 87Sr/86Sr(i) and Nd(t) have identical values within error across the lobe, except in the central leucogranite, the most silica rich phase, which shows a slightly more crustal signature. Field, structural, geochemical and isotopic data suggest that fractionation was the dominant process causing compositional variation in this lobe. It is envisioned that this fractionation/crystal sorting occurred in a vertically ﬂowing and evolving magma column with the present map pattern representing a cross-section of this column. Thus the areal extent of the lobe represents a minimum size of interconnected melt at the emplacement level of the Tuolumne Batholith and, given its marginal position, limited width and proximity to colder host rocks, implies that fractionation in larger chambers likely occurred in the main Tuolumne Batholith magma chamber(s). KEY WORDS:

fractional crystallisation, magmatic processes, plutons

The w1100 km2 Tuolumne Batholith in the central Sierra Nevada, California (Fig. 1), represents a spectacularly exposed, protracted record of plutonic assembly and internal diﬀerentiation in a large, open system continental arc magma chamber. Controversies over the assembly and internal diﬀerentiation processes in the Tuolumne Batholith persist, despite increasingly large geochronological and geochemical data sets collected from the batholith over the past 25 years. Central to this debate is the long-standing question of whether the compositional diversity preserved in the Tuolumne Batholith largely reﬂects source region heterogeneity, contamination, internal fractional crystallisation, or a combination of processes. Bateman & Chappell (1979) used geochemical data to argue that compositional diversity in the Tuolumne Batholith formed largely by in situ fractional crystallisation of a single, batholith-sized pulse. Kistler et al. (1986) used isotopic data to argue that much of the compositional diversity reﬂects mixing between two separate mantle and crustal derived melts. More recently, Coleman et al. (2004) and Glazner et al. (2004)

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016065

argued that a 94–85 Ma intrusion interval requires growth of the Tuolumne Batholith by numerous small increments of magma, although the speciﬁc size and longevity of these increments remains contentious. In this model, compositional diversity and elemental variability in the batholith largely reﬂect variations in partial melt fraction and/or melt source heterogeneity. Based on ﬁeld studies and published geochemistry, Paterson & Vernon (1995) Z { a´k & Paterson (2006) and Memeti et al. (2007a) argued that the diversity was caused by the intrusion of large, nested diapirs derived from diﬀerent source compositions and that fractional crystallisation occurred within these pulses, and mixing/mingling occurred along the margins between major pulses. The size and longevity of the pulses proposed by these authors was also not speciﬁed. The above also highlights the issue of whether large magma chambers commonly exist in magmatic arcs and the size and number of pulses juxtaposed during batholith construction (Pitcher & Berger 1972; Hutton 1992; Lagarde et al. 1990;

174

R. C. ECONOMOS ET AL.

Figure 1 Map of the Half Dome granodiorite lobe and sample locations, with location maps of Yosemite National Park in CA and inset map of the Tuolumne Batholith. Main TIS units are labelled: Kuna Crest (KC); Half Dome (HD); Cathedral Peak (CP); Johnson Porphyry (JP). Lobe units are labelled: equigranular Half Dome granodiorite (eHd); porphyritic Half Dome granodiorite (pHd), leucocratic Half Dome granite to granodiorite (lHd); and leucogranite (lg).

Paterson & Miller 1998; Wiebe & Collins 1998; Johnson et al. 1999; Miller & Paterson 2001; Coleman et al. 2004; Matzel et al. 2005, 2007; Memeti et al. 2007b). It is suggested that a useful approach to address these issues is to examine magmatic lobes extending out from the main batholith (Memeti et al. 2005). Peripheral lobes in general may oﬀer a relatively simple and short history, and may be more readily amenable to petrogenetic interpretation, in comparison to the larger domains that are complicated by reheating and juxtaposition of later pulses. (Memeti et al. 2005; Economos et al. 2005; Memeti et al. 2007a).

1. Geologic setting The Tuolumne Batholith, emplaced at a depth of 6–10 km (Ague & Brimhall 1988; Webber et al. 2001; Gray 2003), is normally zoned from outer maﬁc units to inner, progressively more felsic units (Fig. 1) (Bateman 1992). The outer units include the 95–92 Ma Glen Aulin and Glacier Point tonalites and Kuna Crest granodiorite (Bateman 1992; Kistler & Fleck 1994; Coleman et al. 2004). Within these is the 91–88 Ma more siliceous Half Dome granodiorite which contains an outer equigranular unit and inner porphyritic unit with K-feldspar

COMPOSITIONAL DIVERSITY, HALF DOME GRANODIORITE

175

Figure 2 Photographs of Half Dome Lobe units: (a) Equigranular Half Dome granodiorite; (b) Porphyritic Half Dome granodiorite; (c) leucocratic porphyritic Half Dome granodiorite; (d) central leucogranite.

phenocrysts up to 3–4 cm in length (Bateman 1992; Kistler & Fleck 1994; Coleman et al. 2004). Further inward is the most aerially extensive unit in the batholith, the 88–85 Ma megacrystic Cathedral Peak granodiorite (Bateman 1992; Kistler and Fleck 1994; Coleman et al. 2004, Matzel et al. 2005). The small central phase is the w87 Ma Johnson Granite Porphyry containing local Cathedral Peak inclusions and K-feldspar phenocrysts (Bateman 1992; Titus et al. 2005; Bracciali 2008). This paper focuses on the Half Dome granodiorite. The most distinctive petrographic characteristics of this unit include approximately equal amounts of large (often >1 cm long) euhedral hornblendes and small subhedral biotites and prominent titanite crystals (Bateman 1992). The Half Dome unit dominates the southwestern portion of the main Tuolumne Batholith and has narrow northward and southward protruding lobes (Fig. 1 inset). The southern lobe was originally mapped as Half Dome granodiorite (Peck 1980; Huber et al. 1989), but was later assigned to the porphyritic Half Dome phase (Bateman 1992). The lobe intrudes the w98 Ma Red Devil Lake granodiorite (Tobisch et al. 1995), the Turner Lake granite, and the Cony Craigs Porphyry (Bateman 1992). The southern tip of the lobe extends into the 97–98 Ma Jackass Lakes pluton (McNulty et al. 1996). Thermal modelling and 40Ar/39Ar thermochronology indicate that the Half Dome lobe may have cooled in a few 105 years (Paterson et al. 2007), whether emplaced as a single or multiple pulses. An ongoing study of detailed U/Pb zircon geochronology in the lobe yields a 206 Pb/238 U ages of 90·120·16 Ma (MSWD of 2·3) for the oldest phase (Fig. 1, ‘pHd’) and 89·680·24 Ma for the youngest phase (Fig.1, ‘lg’) (Memeti et al. 2007c; Memeti et al. in press).

2. Results 2.1. Field relationships The southern Half Dome lobe has been remapped at a 1:10 000 scale. The new mapping shows that the lobe can be divided into four compositionally and texturally distinct phases (Figs 1, 2). The outermost phase is a granodiorite to quartz monzodiorite that is generally equivalent to the equigranular variety of the Half Dome granodiorite in the main batholith (hereafter ‘outer phase’ [eHd]). It is medium-grained and contains up to 2 cm euhedral hornblende and up to 1 cm euhedral titanite (Figs 1, 2a). This phase grades inwards into a biotitedominated, porphyritic granodiorite containing K-feldspar megacrysts up to 5 cm in length (hereafter, ‘porphyritic phase’ [pHd]) (Figs 1, 2b). This phase grades inwards to a more leucocratic, medium-grained biotite granodiorite to monzogranite with 0–2% modal hornblende, 3–5% modal biotite, and sparse, 1–3 cm K-feldspar megacrysts (hereafter, central phase [lHd]) (Figs 1, 2c). The innermost central phase is a ﬁne-grained monzogranite dominated by quartz and K-feldspar with less than 1% modal biotite (hereafter, ‘leucogranite’ [lg]) (Figs 1, 2d). These units are organised in a symmetrical map pattern with NW-striking contacts aligned parallel to the margins of the lobe (Fig. 1); textural and compositional characteristics of units on the east and west sides of the lobe are identical. Whilst the compositions within each phase are relatively homogeneous, contacts between the outer, porphyritic, and central phases are gradational over zones 50–60 m wide. As such, contact zones were characterised by gradual changes in

176

R. C. ECONOMOS ET AL.

modal mineralogy, microstructure and colour index, and deﬁned in the ﬁeld as the regions of largest gradients of the above. These gradational contacts remain subvertical across fairly signiﬁcant changes in topography, and are thus interpreted to be steeply dipping everywhere within the lobe. Cross-cutting dikes from diﬀerent phases reveal that the outer phase is the oldest, and that phases young towards the centre of the lobe. The contact between the leucogranite and its surrounding central phase is unusual in the lobe in that it is sharp in most localities and has both shallow and steep dips. The contact between the lobe and its older plutonic host rocks varies dramatically in structure and composition along strike. The tip of the lobe exhibits a highly irregular pattern and is also structurally complex. Compositions range from granite to monzodiorite, with predominantly granodiorite compositions. This tip area contains many randomly oriented blocks of host rock granitic units and rhyolitic to dacitic metatuﬀs (observed only as blocks) ranging from centimetreto tens of metres-scales. Enclave swarms tens of metres in diameter that contain hundreds of maﬁc enclaves are common. Dikes that cross cut host rocks extrude southward away from the very southern tip of the lobe. The southern lobe contact north of the tip is characterised by a 100–200 m-wide structurally complex zone between the equigranular Half Dome granodiorite and the Turner Lake granite (Fig. 1, ‘Turner Lake/Half Dome mingled zone’). Common complexities along this contact include mingling between distinctive hornblende-free Turner Lake granite and Half Dome lobe granodiorite, blocks of Turner Lake granite surrounded by Half Dome granodiorite, and abundant schlieren. The northern lobe–host rock contact is generally sharp. Magmatic fabrics in the lobe include a wNNW–SSE foliation and an wE–W foliation that overprints all gradational internal contacts in the lobe (Fig. 1).

2.2. Petrography The outer phase of the Half Dome lobe consists of 53–61% plagioclase, 12–22% quartz, 6% K-feldspar, 10% biotite and 7% hornblende. Accessory minerals include abundant 1·5– 3 mm euhedral titanite, apatite, magnetite and zircon. Plagioclase, hornblende and biotite are sub- to euhedral. K-feldspars are large and anhedral, commonly showing perthite exsolution and appear to occupy interstices between more euhedral phases. Plagioclase and hornblende inclusions also commonly occur in K-feldspars. Myrmekite occurs in this phase, and chloritisation of biotites, undulose extinction in quartz and mild deformation of plagioclase twins are also observed. The porphyritic phase contains 43–54% plagioclase, 20–28% quartz, 14% K-feldspar, 8% biotite and 3% hornblende. Accessory minerals include titanite, apatite, magnetite, zircon and allanite. Titanite is less abundant, but still large and euhedral. K-feldspars have a similar morphology to the outer phase and also show perthite texture, but are occasionally much larger, up to 1·5 cm in length. The central phase contains 38% plagioclase, 28–32% quartz, 25–29% K-feldspar, 2–5% biotite, and 0–2% hornblende. Titanite and magnetite are present, but less abundant in this phase. K-feldspars are subhedral and perthitic. Whilst K-feldspars are still poikiolitic, they no longer display the interstitial texture from outer phases. Zircon and apatite are also common accessory phases. The leucogranite phase contains 35% plagioclase, 48% quartz, 16% K-feldspar, and 1% biotite. Titanite is far less abundant than in other phases and is subhedral. Grain sizes in the leucogranite phase are equigranular and markedly smaller than in other phases. Alteration is much more prevalent in this phase, including complete chloritisation of biotite and strong

sericitisation of plagioclase crystals. Minor sulﬁdes also occur in this unit. In summary, modal mineralogy in the lobe shows a clear increase in quartz content and a decrease in plagioclase, hornblende, biotite and titanite toward the centre of the lobe. Total K-feldspar content increases inward and peaks in the megacryst bearing phase.

2.3. Geochemistry Seven samples for geochemical analyses were collected along a N–S transect across the dominantly NE–SW zoning in the lobe (Fig. 1, Table 1). Samples representative of the main units were selected for analyses. Whole rock major and trace element data were collected by X-ray ﬂourescence (XRF) and inductively coupled plasma mass spectrometer (ICP-MS) at the Geoanalytical laboratory at Washington State University. Sample chips were selected and ground in a tungsten carbide mill, mixed with dilithium tetraborate, and fused in graphite crucibles for major element analysis and some minor elements. REE analyses of low abundance trace elements were analysed on a quadrupole mass spectrometer with an inductively coupled argon plasma source. Detection limits were at or below chondrite levels. Samples for ICP-MS analysis were ground in an iron bowl in a shatterbox swing mill. Two grams of rock powder were mixed with an equal amount of lithium tetraborate ﬂux, placed into a carbon crucible and fused. The resulting fused bead was re-ground and dissolved in a mixture of HF, HNO3, and HCLO4. The samples were then dried and re-dissolved and mixed with a standard of In, Re and Ru, used to correct for instrument drift. Measurements were conducted on a Sciex Elan model 250 ICP-MS. Calibration curves were constructed from common silicate rock standards run with the samples for each element and unknown concentrations were computed from these curves. Whole-rock Nd and Sr isotopic data were also collected for each phase of the lobe (Fig. 3e, f, Table 2). Two hundred mg of rock powder for each sample was dissolved in a mixture of HF and HNO3 and spiked with mixed 150Nd-147Sm spike in a sealed Teﬂon bomb at 180(C for 5–7 days. Separation of Rb, Sr and the REE group followed standard cation-exchange procedures. Separation of Sm and Nd used alpha-HIBA on cation exchange resin. Strontium was separated using Sr-spec column chemistry and HNO3. Neodymium was loaded on single Re ﬁlaments with dilute HCl, and Sm were loaded on single-Ta ﬁlaments with H3PO4. Strontium was loaded with H3PO4 and TaCl5 emitter on a single Re ﬁlament. All analyses were performed on the Micromass Sector-54 mass spectrometer at the University of North Carolina. Neodymium was analysed in dynamic multicollector mode as NdO using an oxygen bleed valve at 1V, and Sm was analysed in static multicollector mode with 147Sm=200 mV. Strontium was analysed in dynamic multicollector mode with 88Sr=3V. Neodymium data are normalised to 146Nd/144Nd=0·7219. Strontium data are normalised to 86Sr/88Sr=0·1194. During the period of analysis, replication of SRM-987 yielded 87Sr/ 86 Sr=0·7102460·000010. Nd data are referenced to La Jolla Nd (143Nd/144Nd=0·511853). Replicate analysis of the UNC J-Nd standard during the period of analysis gave 143Nd/ 144 Nd=0·5120990·000005. This internal standard is also referenced to LaJolla Nd but is run more frequently. No bias correction has been applied based on repeated measurements of both standards. Initial Sr ratios were calculated using 87Rb=1·4210 11 yr, and using Rb and Sr concentrations measured by ICP-MS. Initial ratios are calculated using a nominal age of 92 Ma for the Half Dome granodiorite (e.g. Coleman et al. 2004).

SiO2

1 2 3 4 5 6 7

Sample

1 2 3 4 5 6 7

1·1 0·7 0·3 0·3 0·5 0·7 1

Yb (ppm)

32 25·1 15·7 17·1 22·5 26·4 29·3

La (ppm)

26 26 16 <10 23 26 19

1 2 3 4 5 6 7

Sample

Li (ppm)

Sample

1 58·22 2 67·2 3 74·9 4 76·17 5 68·74 6 66·68 7 61·7 ICP-MS measurements:

Sample

XRF measurements in wt%:

3 4 3 3 2 4 3

Hf (ppm)

63·5 46·1 18·9 17·1 40·5 48 59·2

Ce (ppm)

1·52 0·62 0·14 0·06 0·48 0·66 1·28

Mg (%)

17·97 15·88 12·9 12·86 15·44 15·9 16·17

AI2O3

13 52 24 67 17 15 13

Pb (ppm)

7·36 5·29 1·55 1·24 4·92 5·59 7·25

Pr (ppm)

0·14 0·09 0·02 <0·01 0·07 0·08 0·12

P (%)

5·78 3·37 1·19 0·88 2·99 3·37 4·93

CaO

11·3 11·8 19·4 21 17·1 13·3 11·8

Th (ppm)

27·5 19·6 4·7 3·4 17·5 20·6 27·2

Nd (ppm)

114 137 182 181 127 150 134

Cr (ppm)

2·4 0·99 0·25 0·11 0·78 1·02 1·98

MgO

7·92 3·84 6·12 6·52 4·02 3·68 4·66

U (ppm)

4·8 3 0·7 0·5 2·7 3·4 4·5

Sm (ppm)

4·16 2·28 0·92 0·78 1·9 2·42 3·84

Fe (%)

4·17 4·17 3·28 3·57 3·97 4·04 3·73

Na2O

1·24 0·86 0·18 0·08 0·77 0·85 1·11

Eu (ppm)

10 9 8 <5 7 9 13

Ni (ppm)

2·46 3·36 4·57 4·49 3·53 3·15 2·81

K2O

4·28 2·68 0·58 0·47 2·26 2·74 3·83

Gd (ppm)

93·9 130·5 192·9 205·4 139·8 118·1 86·2

Rb (ppm)

6·26 3·37 1·35 1·14 2·87 3·51 5·57

Fe2O3

0·58 0·37 0·08 0·06 0·29 0·36 0·49

Tb (ppm)

872·4 697·2 151·4 72·6 624·6 679·1 797·2

Sr (ppm)

0·11 0·07 0·02 0·02 0·06 0·07 0·09

MnO

2·4 1·5 0·35 0·31 1·16 1·47 2·1

Dy (ppm)

12·6 7·2 2·4 2·2 6·2 7·9 11·3

Y (ppm)

0·87 0·48 0·14 0·09 0·43 0·49 0·75

TiO2

0·47 0·27 0·08 0·07 0·21 0·28 0·39

Ho (ppm)

77·5 124·8 60·5 56·8 73·2 138·9 115·1

Zr (ppm)

0·29 0·18 0·04 0·01 0·15 0·17 0·27

P2O5

1·21 0·78 0·2 0·25 0·59 0·79 1·23

Er (ppm)

9 7 5 7 7 7 9

Nb (ppm)

0·02 0·02 0·03 0·03 0·02 0·02 0·02

Cr2O3

0·17 0·1 <0·05 <0·05 0·08 0·12 0·16

Tm (ppm)

3·3 5·7 6·4 3·9 6·7 4·7 3·8

Cs (ppm)

0·45 0·25 0·3 0·1 0·4 0·45 0·4

LOI

0·22 0·13 0·08 0·06 0·12 0·13 0·17

Lu (ppm)

889·6 1119·6 194 45·5 825·4 926·5 1107·3

Ba (ppm)

98·99 99·32 98·97 99·47 99·38 98·86 98·42

Sum

Table 1 Major oxide and trace element data collected at the Geoanalytic Laboratory at Washington State University. Major oxides were measured by XRF and trace elements were measured by ICP-MS.

COMPOSITIONAL DIVERSITY, HALF DOME GRANODIORITE 177

178

R. C. ECONOMOS ET AL.

Figure 3 (a) Trace elements Ba, Sr, Rb, Zr and SiO2 plotted across the lobe to demonstrate symmetrical pattern; (b) The same trace elements plotted against SiO2; (c) Sr vs. Rb for the Half Dome lobe and the entire Tuolumne Batholith (Bateman & Chappell 1979; Bateman 1992; Gray 2003); (d) Rare Earth Elements normalised to chondrite (Boynton 1984); (e) (f) Sr(i) and Nd vs. SiO2 for the Tuolumne Batholith (Gray 2003; Kistler et al. 1986) and the Half Dome lobe.

The compositional and textural symmetry of the lobe is strongly reﬂected in geochemical trends. SiO2 content across the lobe displays a symmetrical inward increase from w60 wt% SiO2 in the outer phase to 76 wt% SiO2 in the leucogranite phase (Fig. 3a). This SiO2 range is impressive, since it is as large as that of all four mapped units of the main batholith combined (Bateman & Chappell 1979). All major oxide trends in both the lobe and the main batholith show fairly linear relationships with SiO2 with very little scatter (Bateman & Chappell 1979). CIPW norm calculations (including hornblende) conﬁrm the mineralogical trends identiﬁed through point counting, including an increase in K-feldspar in the porphyritic phase of the lobe.

Trace element analyses reveal a symmetrical increase in Rb content towards the centre of the lobe, whereas Ba and Sr decrease toward the centre of the lobe with a major drop oﬀ in the leucogranite phase (Fig. 3a). When plotted against SiO2, Rb and Sr show curvilinear trends, while Ba is more scattered (Fig. 3b). These trace element variations are unlike trends in the main batholith, which tend to show weak or no correlations with SiO2 (Kistler et al. 1986; Fig. 2). Zr displays a more complex trend, increasing from the outer phase to the porphyritic phase and then dropping steeply across the gradational contact between the porphyritic phase and the central phase (Figs 3a, 4a, b). All REE concentrations decrease, with increasing SiO2 toward the centre of the lobe. Light REEs are more

COMPOSITIONAL DIVERSITY, HALF DOME GRANODIORITE 87

86

143

179

144

Table 2 Sr/ Sr and Nd/ Nd are measured ratios with 2-sigma analytical errors in parentheses. Initial Sr and Nd ratios are calculated using an age of 92 Ma. Epsilon Nd values at the crystallisation age are calculated using 143Nd/144Nd(CHUR, 0 Ma) =0·512638 and 147Sm/144Nd(CHUR, 0 Ma) = 0·1967. Rb and Sr concentration data determined by ICP-MS and Sm and Nd concentration data determined by isotope dilution. Errors in Rb and Sr are within 1% and 0·1% respectively of the amount present. Errors for Sm and Nd are within 2% of the amount present and epsilon values calculated using Sm/Nd from ICP data are within 5% of those calculated by ID-TIMS. Based on these considerations and replication of standards and rocks (e.g. BCR-1; see Miller et al. 2000), conservative errors of 0·00005 are assigned to initial 87Sr/86Sr and 0·5 epsilon units to initial Nd values. Sample 3 4 5 6 7

Rb ppm

Sr ppm

193 205 138 118 86

151 73 623 679 797

87

Sr/86Sr

0·711382 (16) 0·716378 (14) 0·707197 (16) 0·707068 (16) 0·706802 (16)

87

Sr/86Sr (92) 0·706556 0·705776 0·706361 0·706412 0·706395

Sm ppm

Nd ppm

0·841 0·427 3·35 3·21 5·27

5·642 2·975 20·62 22·89 30·86

enriched than heavy REEs in all samples, and light REE enrichment relative to heavy REEs becomes more pronounced in the central phases (Fig. 3d). Middle REEs show a decrease, particularly in the core aplite, and have a ‘scooped’ shape that is characteristic of other titanite-bearing Sierran aplites (Glazner et al. 2008). Initial Sr and Nd isotopic compositions (and Nd(t)) are invariant within error for the outer, porphyritic and central phases (spanning 15 wt% silica) with 87Sr/86Sr(i) =0·7064 and Nd(t) = 4·6 (Fig. 3e, f). The leucogranite unit has 87Sr/ 86 Sr(i) =0·7058 and Nd(t) = 5·8, and is therefore isotopically distinct from the other units.

3. Discussion A robust petrogenetic model for the southern Half Dome lobe must address: (1) normal zoning from maﬁc/intermediate to felsic that is symmetrical from margin to centre; (2) gradational, steeply dipping magmatic contacts between all units and sharp and variably dipping contacts with the leucogranite; (3) hornblende-rich ‘cumulate-like’ zones and local layering and associated maﬁc enclave swarms at the margin of the lobe; (4) smooth decreases in plagioclase and maﬁc mineral content and appearance of K-feldspar phenocrysts across a narrow gradational zone; (5) smooth and linear major element variation (excluding Na2O) with SiO2; (6) regular trace element variation with SiO2 and distance for many trace elements (Fig. 3a, b); and (7) invariant Nd and Sr isotopic composition across a wide range of SiO2 and with distance from the lobe margins with a minor variation in the core leucogranite (Fig. 3e, f). The symmetric zoning pattern is one of the key compositional aspects of the Half Dome lobe that distinguishes it from the Half Dome elsewhere in the Tuolumne. For example, the much larger mass of the Half Dome granodiorite that comprises most of the southwestern exposure of the Tuolumne (see Fig. 1 inset) is quite homogeneous in mineralogy and major element composition (Gray 2003). In addition, the contacts in the Half Dome lobe are everywhere gradational and are thus markedly simpler than those between the major mapped units in the main batholith, which may change from gradational to sharp along strike (e.g. Bateman 1992). Compositional zoning of plutons has been attributed to a wide variety of processes including: crystal-liquid fractionation, mixing of compositionally diﬀerent melts, restite unmixing, and emplacement of compositionally distinct pulses at a more or less constant locus of intrusion (cf. Ragland & Butler 1972; Bateman & Nokelberg 1978; Bateman & Chappell 1979; Halliday et al. 1980; Reid et al. 1983; Scambos et al. 1986; Hill et al. 1988; Sawka et al. 1990; Walawender et al. 1990; Barbey et al. 2001; Wyborn et al. 2001).

147

Sm/144Nd 0·0901 0·0867 0·0981 0·0848 0·1032

143

Nd/144Nd

0·512342 (65) 0·512280 (09) 0·512326 (09) 0·512349 (08) 0·512344 (12)

Nd (92)

TDM (Ma)

4·53 5·70 4·93 4·33 4·63

871 921 950 823 970

Given the ﬁeld and petrologic data presented above it is considered unlikely that the lobe was emplaced with abundant restite and/or that restite unmixing was important (e.g., Miller et al. 2007). Magma mixing can occur between basaltic melts and felsic melts derived by remelting of maﬁc underplated materials (e.g. Coleman et al. 1992; 1995; Ratajeski et al. 2001), and so it cannot be unequivocally ruled out by isotopic data alone. The strikingly linear major element covariation and strong linear correlations with bulk composition are what might be expected for magma mixing, but they could also result from bulk unmixing in the lobe. Magma mixing between the enclave-forming maﬁc magma with any of the more felsic magmas is precluded by trace element trends (e.g. Rb vs. Sr, Fig. 3c), which clearly show that the enclave has too low Sr to serve as an appropriate maﬁc end member. Several trace elements display non-linear or curvilinear co-variations (Fig. 4c, d) that preclude a simple two component mixing model. Another ﬁrst order constraint comes from the whole rock isotopic data, which, for all phases but the central aplite, excludes mixing of melts from two isotopically diﬀerent sources. This suggests a relatively ‘closed-system’ process. Along the margins of the lobe, evidence for the operation of crystal–liquid fractionation processes is indicated by the hornblende and biotite dominated zones and development of schlieren troughs and tubes (e.g. Weinberg et al. 2001; Z { a´k & Paterson 2005). Away from the margins of the lobe, where transitions between phases are gradational, outcrop-scale evidence for these processes are also observed in the form of crystal accumulations and crystal-poor leucocratic veins (Paterson et al. 2005). These rocks have not been analysed, but within the Half Dome lobe they are the ones most obviously aﬀected by accumulation and/or melt segregation. Their relatively minor volumetric proportion suggests that they are unlikely to form a primary cumulate assemblage for the lobe, and such rocks elsewhere in the Tuolumne Batholith cannot be easily related to the main geochemical trends (Reid et al. 1983; Paterson et al. 2008). They nevertheless provide evidence for redistribution of crystals and melt, and they are thus regarded as evidence for the operation of at least local crystal-melt movements in a magma chamber at the site of emplacement of the lobe. All petrographic evidence is also permissive of fractional crystallisation, including increasing quartz and K-feldspar content towards the centre of the lobe and decreasing total plagioclase content. Overgrowth relationships show transitions from clinopyroxene to hornblende to biotite. The leucogranite phase represents a highly evolved melt. A simple least-squares mixing analysis of major element compositions revealed that approximately equal proportions of a residual outer phase and a leucocratic central phase would be produced by fractionating

180

R. C. ECONOMOS ET AL.

Figure 4 (a) Zircon related elements, U, Th, and Zr, plotted across the lobe showing an inﬂection point in the porphyritic unit; (b) the same elements plotted vs. SiO2. (c) (d) Two examples of non-linear trace element trends in the Half Dome lobe: Y vs. Sr and Zr vs. Rb.

the porphyritic phase. However, not enough of the outer phase is present at the level of exposure to account for the fractionation of both the central and leucocratic phase. This is one of several observations that lead the present authors to propose a vertically evolving fractionation system in which some residue resides below the present level of exposure. Trace element variations are also consistent with fractional crystallisation. Incompatible elements like Rb show an increase with SiO2 and distance from the host rock contacts, whilst more compatible elements like Sr show a decrease (Fig. 3a), attributed to fractional crystallisation of the major minerals, such as hornblende and plagioclase. These minerals are found in higher abundances in the more marginal phases of the lobe. The pronounced drops in Sr and Ba observed in the leucogranite phase are controlled by feldspar fractional crystallisation, especially K-feldspar, again conﬁrmed by changes in modal mineralogy (Figs 2, 3). A modest increase in the Ce/Yb ratio is probably indicative of hornblende control until higher SiO2 values are reached in the interior phases, whereupon Ce/Yb drops markedly. This likely reﬂects saturation and removal of phase(s) that strongly concentrate the light to middle rare earth elements, and extraction of melt that has been depleted in these elements. The removal of middle REEs that produces the distinctive ‘scoop’ shape of the rare earth element trend in the leucocratic phases of the lobe is likely attributable to accessory mineral fractional crystallisation, especially titanite (e.g. Glazner et al. 2008). Finally, it is re-emphasised that the isotopic data are a strong indication for dominantly closed-system fractional crystallisation occurring in the Half Dome lobe. Both 87Sr/86Sr(i) and Nd(t) values are equal to each other within error over an impressively large variation in SiO2 content (Fig. 3e, f). This

isotopic dataset bears striking similarity to ﬁeld data, particularly contact relationships. Units that share gradational contacts have identical isotopic values, whilst the single unit with a sharp contact shows a slight isotopic shift. Geochemical modelling to test the above hypotheses is underway. However, such modelling remains fairly non unique without detailed mineral geochemistry, full consideration of both major, trace, REE, and isotopic data and the roles/ volumes of both major and accessory minerals and models designed to test all likely scenarios, and thus is beyond the scope of this paper. Another challenge for geochemical modelling, including modelling of the Tuolumne Batholith, is the increased evidence of recycling and mixing of crystal populations such as now documented for the zircon populations (Matzel et al. 2005; Miller et al. 2007) and established even at the mesoscale (Paterson et al. 2008). Until these modelling studies are complete, the preferred interpretation is that the lobe represents a frozen cross-section of a vertically evolving and fractionating magma system (Fig. 5). This would allow the possibility that not all of the present exposure in the lobe was laterally connected melt at one time; but if not laterally, it was so vertically. Whilst the percentage of crystallisation beyond the formation of a crystal network is diﬃcult to ascertain, modelling by Bachmann & Bergantz (2004) indicates that relatively crystal-poor magmas can be removed from a mush of 40% to >50% crystals. The present authors envision a smaller scale version of the upper portions of models presented by Sisson (2005) and Ratajeski et al. (2005), in which crystal mush zones along conduit walls grade into a more upwardly mobile crystal mush centre. The chemical gradients, implying chemical communication between magma, cumulates at the host rock contact,

COMPOSITIONAL DIVERSITY, HALF DOME GRANODIORITE

181

4. Conclusions

Figure 5 Schematic depicting the major boundary conditions envisioned for a vertically evolving column of fractionating material.

and overprinting magmatic fabrics in this lobe are all timetransgressive (both laterally and vertically); features consistent with this model. Thus, the compositional diversity in the lobe is generated by processes occurring both at and below the level of emplacement, but these processes are linked to each other by interconnected melt in an evolving system. If a similar system operated in the main batholith, in which pulses were disrupted by later intrusions, this nicely explains the diﬀerence in isotopic systematics between the Half Dome lobe and the entire Half Dome unit, with the former being isotopically homogeneous while the latter shows a much more complex history (Kistler et al. 1986; Kistler & Fleck 1994, Memeti et al. 2007b; Matzel et al. 2007). Such a vertically evolving, fractionating system would predict the same or more likely even greater vertical extent than the cross-sectional area of the lobe. If such a volume of interconnected melt is achieved in the thermally less favourable environment of the lobe, larger bodies of interconnected melt should be expected within the main body of the Tuolumne Batholith. This conclusion is supported in the main batholith by magmatic fabrics that overprint all internal contacts (Z { a´k et al. 2007), broad zones of mixing/mingling along these contacts and gradations between them (Bateman 1992; Z { a´k & Paterson 2006; Paterson et al. 2008), and broad zones of nearly identical U/Pb zircon ages and related geochemistry in the large central Cathedral Peak unit (Burgess & Miller 2008). One interesting question is why the major oxide diversity in the lobe is greater than that preserved in any unit in the main batholith. Was this signal somehow removed in the main batholith? Or if never achieved by an individual unit in the main batholith, does it imply that the elevated geothermal gradient and rapid crystallisation predicted in the lobes may produce more rather than less fractional crystallisation in natural systems? The most critical diﬀerences in boundary conditions between the lobe and the main batholith are thermal gradient during crystallisation (Spera & Bohrson 2001; Paterson et al. 2003; Memeti et al. 2007a) and presence of additional distinct magma pulses in the main batholith (Spera & Bohrson 2001). The eﬀects of these boundary condition diﬀerences on the mechanical separation of crystals and liquid warrants additional discussion.

The mapping, structural, chemical and isotopic evidence suggests that a vertically ﬂowing and fractionating column of magma, derived from a largely isotopically homogeneous source, played the dominant role in the formation of compositional diversity in the southern Half Dome granodiorite lobe of the Tuolumne Batholith. This keeps open the possibility that fractional crystallisation may have occurred in the main chamber over length and volume scales signiﬁcantly larger than that of the lobe, requiring bodies of interconnected melt at least as large as the lobe at the time of emplacement of units in the main batholith. These results suggest that fractional crystallisation at or near the level of emplacement of the Tuolumne Batholith is capable of producing major oxide heterogeneity of the magnitude seen in the entire batholith. Why this level of fractional crystallisation is not preserved in the main batholith may be that it is masked by subsequent processes (Memeti et al. 2007b), or that an elevated thermal gradient in the lobe was a driving factor for the degree of fractional crystallisation. A lower thermal gradient and juxtaposition of diﬀerent batches in the main batholith may have allowed mixing to play a more signiﬁcant role.

5. Acknowledgements This work was supported by funds from the University of Southern California Department of Earth Sciences, by NSF grant EAR-0073943 (to Paterson), NSF EAR-0074099 (to Miller) and by Czech Academy of Sciences Grant No. KJB 3111403 (to Z { a´k). Special thanks to J. Lawford Anderson for his intellectual contribution, and to Claire Wilke and Gayle Hough for their assistance in the ﬁeld. Thanks to Dr Calvin Miller and Dr Allen Glazner for their constructive, helpful reviews.

6. References Ague, J. J. & Brimhall, G. H. 1988. Magmatic arc asymmetry and distribution of anomalous plutonic belts in the batholiths of California; eﬀects of assimilation, crustal thickness, and depth of crystallization. Geological Society of America Bulletin 100, 912– 27. Bachmann, O. & Bergantz, G. W. 2004. On the origin of crystal-poor rhyolites extracted from batholithic crystal mushes. Journal of Petrology 45 (8), 1565–82. Barbey, P., Nachit, H. & Pons, J. 2001. Magma–host interactions during diﬀerentiation and emplacement of a shallow-level, zoned granitic pluton Tarc¸ouate pluton, Morocco: implications for magma emplacement. Lithos 58, 125–43. Bateman, P. C. 1992. Plutonism in the central part of the Sierra Nevada Batholith, California. US Geological Survey Professional Paper P 1483. Bateman, P. C. & Chappell, B. W. 1979. Crystallization, fractionation, and solidiﬁcation of the Tuolumne Intrusive Series, Yosemite National Park, California. Geological Society of America Bulletin 90 (5), 1465–82. Bateman, P. C. & Nokelberg, W. J. 1978. Solidiﬁcation of the Mount Givens Granodiorite, Sierra Nevada, California. Journal of Geology 86, 563–79. Boynton, W. V. 1984. Cosmochemistry of the rare earth elements: meteorite studies. In Henderson, P. (ed.) Rare Earth Element Geochemistry, 63–114. Amsterdam: Elsevier. Bracciali, L., Paterson, S. R., Memeti, V., Rocchi, S., Matzel, J. & Mundil, R. 2008. Build-up of the Tuolumne Batholith, California: the Johnson Granite Porphyry. In LASI 3 Conference Extended Abstract Volume, Physical Geology of Subvolcanic Systems: Laccolith, Sills and Dykes. Elba Island, Italy, September 2008. Burgess, S. D. & Miller, J. S. 2008. Construction, solidiﬁcation, and internal diﬀerentiation of a large felsic arc pluton. Cathedral Peak granodiorite, Sierra Nevada Batholith. In Zellmer, G. & Annen,

182

R. C. ECONOMOS ET AL.

C. (eds) Geological Society, London, Special Publication 304, 203–33. Bath, UK: The Geological Society Publishing House. Coleman, D. S., Frost, T. P. & Glazner, A. F. 1992. Evidence from the Lamarck Granodiorite for rapid Late Cretaceous crust formation in California. Science 258, 1924–26. Coleman, D. S., Glazner, A. F., Miller, J. S., Bradford, K. J., Frost, T. P., Joye, J. L. & Bachl, C. A. 1995. Exposure of a Late Cretaceous layered maﬁc-felsic magma system in the central Sierra Nevada batholith, California. Contributions to Mineralogy and Petrology 120, 129–36. Coleman, D. S., Walt, G. & Glazner, A. F. 2004. Rethinking the emplacement and evolution of zoned plutons: Geochronologic evidence for incremental assembly of the Tuolumne Batholith, California. Geology 32, 433–6. Economos, R. C., Miller, J. S., Paterson, S. R., Memeti, V., Erdmann, S. & Z { a´k, J. 2005. The Role of Fractionation at the Emplacement Level in the Tuolumne Batholith, Sierra Nevada, CA. Geological Society of America Abstracts with Programs 37, 554. Glazner, A. F., Bartley, J. M., Coleman, D. S., Gray, W. & Taylor, R. Z. 2004. Are plutons assembled over millions of years by amalgamation from small magma chambers? GSA Today 14, 4–11. Glazner, A. F., Coleman, D. S. & Bartley, J. M. 2008. The tenuous connection between high-silica rhyolites and granodiorite plutons. Geology 36, 183–6. Gray, W. 2003. Chemical and thermal evolution of the Late Cretaceous Tuolumne Batholith, Yosemite National Park, California. Ph.D. Thesis. Chapel Hill, University of North Carolina. 202 pp. Halliday, A. N., Stephens, W. E. & Harmon, R. S. 1980. Rb–Sr and O isotopic relationships in three zoned Caledonian granitic plutons, Southern Uplands, Scotland: evidence for varied sources and hybridisation of magmas. Journal of the Geological Society, London 137, 329–48. Hill, M., O’Neil, J. R., Noyes, H., Frey, F. A. & Wones, D. R. 1988. Sr, Nd and O isotope variations in compositionally zoned and unzoned plutons in the central Sierra Nevada Batholith. American Journal of Science 288–A, 213–41. Huber, N. K., Bateman, P. C. & Wahrhaftig, C. 1989. Geologic map of Yosemite National Park and vicinity, California. US Geological Survey Report I-1874. Scale 1:125,000, 1 sheet. Hutton, D. H. W. 1992. Granite sheeted complexes: evidence for the dyking ascent mechanism. Transactions of the Royal Society of Edinburgh: Earth Sciences 83, 377–82. Johnson, S. E., Paterson, S. R. & Tate, M. C. 1999. Structure and emplacement history of multiple-center, cone-sheet-bearing ring complex: the Zarza Intrusive Complex, Baja California, Mexico. Geological Society of America Bulletin 111, 607–19. Kistler, R. W., Chappell, B. W., Peck, D. L. & Bateman, P. C. 1986. Isotopic variation in the Tuolumne Batholith, central Sierra Nevada: California. Contributions to Mineralogy and Petrology 94, 205–20. Kistler, R. W. & Fleck, R. J. 1994. Field guide for a transect of the central Sierra Nevada, California: Geochronology and isotope geology. US Geological Survey Open-File Report 94-267, 51 pp. Lagarde, J. L., Brun, J. P. & Gapais, D. 1990. Formation of epizonal granitic plutons by in situ assemblage of laterally expanding magma. Comptes Rendus de l’Acade´mie des Sciences Paris 130, Serie II, 1109–14. Matzel, J., Mundil, R., Paterson, S. R., Renne, P. & Nomade, S. 2005. Evaluating pluton growth models using high resolution geochronology: Tuolumne Batholith, Sierra Nevada, CA. Geological Society of America Abstracts with Programs 37 (7), 131. Matzel, J., Mundil, R., Miller, J. S., Wooden, J., Mazdab, F., Paterson, S. & Memeti, V. 2007. Zircon growth and recycling during the assembly of the Tuolumne batholith, Sierra Nevada, California. Goldschmidt Conference Abstracts 2007, A639. McNulty, B. A., Tong, W. & Tobisch, O. T. 1996. Assembly of a dike-fed magma chamber; the Jackass Lakes Pluton, central Sierra Nevada, California. GSA Bulletin 108, 926–40. Memeti, V., Paterson, S. R., Economos, R. C., Z { a´k, J. & Erdmann, S. 2005. Deciphering Chamber Growth and Internal Magma Chamber Processes Using Magmatic Lobes as Snapshots of Processes During the Construction of the Tuolumne Batholith, Sierra Nevada. Geological Society of America Abstracts with Programs 37 (7), 554. Memeti, V., Paterson, S. R., Matzel, J. & Mundil, R. 2007a. Magmatic lobes as windows into long lived magma chambers: Tuolumne batholith, Sierra Nevada, California. In Miller, J. A. & Kisters, A. F. M. (eds) 6th International Hutton Symposium Abstract Volume and Program Guide. Stellenbosch, South Africa: Department of

Geology, Geography and Environmental Sciences, University of Stellenbosch. 236 pp. Memeti, V., Paterson, S., Matzel, J., Mundil, R., Ducea, M. & Miller, J. 2007b. Dynamics of a magma chamber: Insights into time and length scales of internal processes in the Tuolumne batholith, CA. Eos Transactions AGU 88 (52). Fall Meeting Supplement. Abstract V42C-07. Memeti, V., Paterson, S. R., Matzel, J., Mundil, R. & Okaya, D. In press. Magmatic lobes as ‘‘snapshots’’ of magma chamber growth and evolution in large, composite batholiths: an example from the Tuolumne Intrusion, Sierra Nevada, CA. Geological Society of America Bulletin. Miller, J. S., Glazner, A. F., Farmer, L., Suayah, I. B. & Kieth, L. 2000. A Sr, Nd, and Pb isotopic study of mantle domains and crustal structure from Miocene volcanic rocks in the Mojave Desert, California. Geological Society of America Bulletin 112(8), 1264–79. Miller, J. S., Matzel, J. E. P., Miller, C. F., Burgess, S. D. & Miller, R. B. 2007. Zircon growth and recycling during the assembly of large, composite arc batholiths. Journal of Volcanology and Geothermal Research 167, 282–99. Miller, R. B. & Paterson, S. R. 2001. Construction of mid-crustal sheeted plutons: Examples from the north Cascades, Washington. Geological Society of America Bulletin 113, 1423–42. Paterson, S. R., Okaya, D. & Z { a´k, J. 2003, Thermal histories of episodically constructed, large volume magma chambers: implications for processes along internal contacts: Geological Society of America Abstracts with Programs 35(6), 93. Paterson, S. R., Vernon, R. H. & Z { a´k, J. 2005. Mechanical Instabilities and Physical Accumulation of K-feldspar Megacrysts in Granitic Magma, Tuolumne Batholith, California, USA. Journal of the Virtual Explorer 18. Paterson, S. R., Okaya, D., Matzel, J., Memeti, V., Mundil, R. & Nomade, S. 2007. Size and longevity of magma chambers in the Tuolumne Batholith: A comparison of thermal modeling and cooling thermochronology. Eos Transactions AGU 88 (52). Fall Meeting Supplement. Abstract V42C-02. Paterson, S. R., Z { a´k, J. & Janosˇek, V. 2008. Growth of complex sheeted zones during recycling of older magmatic units into younger: Sawmill Canyon area, Tuolumne batholith, Sierra Nevada, California. Journal of Volcanology and Geothermal Research 177 (2), 457–84. Paterson, S. R. & Miller, R. B. 1998. Mid-crustal magmatic sheets in the Cascades Mountains, Washington: implications for magma ascent. Journal of Structural Geology 20, 1345–63. Paterson, S. R. & Vernon, R. 1995. Bursting the bubble of ballooning plutons; a return to nested diapirs emplaced by multiple processes. Geological Society of America Bulletin 107, 1356–80. Peck, D. L. 1980. Geologic map of the Merced Peak Quadrangle; central Sierra Nevada, California. US Geological Survey Report GQ-1531. Scale 1:62,500. 1 sheet. Pitcher, W. S. & Berger, A. R. 1972. The Geology of Donegal: a Study of Granite Emplacement and Unroofing. New York: Wiley. 435 pp. Ragland, P. C. & Butler, J. R. 1972. Crystallization of the West Farrington Pluton, North Carolina, USA. Journal of Petrology 13, 381–404. Ratajeski, K., Glazner, A. F. & Miller, B. V. 2001. Geology and geochemistry of maﬁc to felsic plutonic rocks in the Cretaceous intrusive suite of Yosemite Valley, California. Geological Society of America Bulletin 113(11), 1486–502. Ratajeski, K., Sisson, T. W. & Glazner, A. F. 2005. Experimental and geochemical evidence for derivation of the El Capitan Granite, California, by partial melting of hydrous gabbroic lower crust. Contributions to Mineralogy and Petrology 149, 713–34. Reid, J. B., Evans, O. C. & Fates, D. G. 1983. Magma mixing in granitic rocks of the central Sierra Nevada, California. Earth and Planetary Science Letters 66, 243–61. Sawka, W. N., Chappell, B. W. & Kistler, R. W. 1990. Granitoid compositional zoning by side-wall boundary layer diﬀerentiation: evidence from the Palisade Crest Batholith, Central Sierra Nevada, California. Journal of Petrology 31, 519–53. Scambos, T. A., Loiselle, M. C. & Wones, D. R. 1986. The Center Pond Pluton; the restite of the story (phase separation and melt evolution in granitoid genesis). American Journal of Science 286, 241–80. Sisson, T. W. 2005. Solidiﬁcation, Zoning, and Homogenization in Sierran Plutons. Geological Society of America Abstracts with Programs 37 (4), 39. Spera, F. J. & Bohrson, W. A. 2001. Energy-Constrained Open-System Magmatic Processes 1: General Model and Energy-Constrained

COMPOSITIONAL DIVERSITY, HALF DOME GRANODIORITE Assimilation and Fractional Crystallization (EC-AFC) Formulation. Journal of Petrology 42, 999–1018. Titus, S. J., Clark, R. & Tikoﬀ, B. 2005. Geologic and geophysical investigation of two ﬁne-grained granites, Sierra Nevada Batholith, California: Evidence for structural controls on emplacement and volcanism. Geological Society of America Bulletin 117 (9), 1256–71. Tobisch, O. T., Saleeby, J. B., Renne, P. R., McNulty, B. A. & Tong, W. 1995. Variations in deformation ﬁelds during development of a large-volume magmatic arc, central Sierra Nevada, California. Geological Society of America Bulletin 107, 148–66. Walawender, M. J., Gastil, R.G., Clinkenbeard, J. P., McCormick, W. V., Eastman, B. G., Wernicke, R. S., Wardlaw, M. S., Gunn, S. H. & Smith, B. M. 1990. Origin and evolution of the zoned La Posta-Type plutons, eastern Peninsular Ranges batholith, southern and Baja California. In Anderson, J. L. (ed.) The Nature and Origin of Cordilleran Magmatism. Geological Society of America Memoir 174, 1–18. Webber, C. E., Candela, P. A., Piccoli, P. M. & Simon, A. C. 2001. Generation of granitic dikes: can texture, mineralogy, and geochemistry be used as guides to determine the mechanism of diking? Geological Society of America Abstracts with Programs 33, A-138. Weinberg, R. F., Sial, A. N. & Pessoa, R. R. 2001. Magma ﬂow within the Tavares pluton, northeastern Brazil: Compositional and ther-

183

mal convection. Geological Society of America Bulletin 113, 508–20. Wiebe, R. A. & Collins, W. J. 1998. Depositional features and stratigraphic sections in granitic plutons: Implications for the emplacement and crystallization of granitic magma. Journal of Structural Geology 20, 1273–89. Wyborn, D., Chappell, B. W. & James, M. 2001. Examples of convective fractionation in high-temperature granites from the Lachlan Fold Belt. Australian Journal of Earth Sciences 48, 531–41. Z { a´k, J., Paterson, S. R. & Memeti, V. 2007. Four magmatic fabrics in the Tuolumne batholith, central Sierra Nevada, California (USA): implications for interpreting fabric patterns in plutons and evolution of magma chambers in the upper crust. GSA Bulletin 119, 194–201. Z { a´k, J. & Paterson, S. R. 2005. Characteristics of internal contacts in the Tuolumne batholith, central Sierra Nevada, California (USA): Implications for episodic emplacement and physical processes in a continental arc magma chamber. Geological Society of America Bulletin 117, 1242–255. Z { a´k, J. & Paterson, S. R. 2006. Roof and walls of the Red Mountain Creek pluton, eastern Sierra Nevada, California (USA): implications for process zones during pluton emplacement. Journal of Structural Geology 28, 575–87.

MS received 18 February 2008. Accepted for publication 18 September 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 185–204, 2010 (for 2009)

The evolution of late-Hercynian granites and rhyolites documented by quartz – a review Axel Mu¨ller1, Alfons M. van den Kerkhof2, Hans-Ju¨rgen Behr2, Andreas Kronz2 and Monika Koch-Mu¨ller3 1

Geological Survey of Norway, N-7491 Trondheim, Norway Email: [email protected]

2

Geowissenschaftliches Zentrum Go¨ttingen, Goldschmidtstr. 3, D-37073 Go¨ttingen, Germany

3

GeoForschungsZentrum Potsdam, Telegrafenberg, D-14473 Potsdam, Germany

ABSTRACT: The potential of igneous quartz for providing a better understanding of magmatic processes is demonstrated by studying late-Hercynian rhyolites and granites from central and western Europe. Cathodoluminescence (CL) reveals growth patterns and alteration structures within igneous quartz reﬂecting the magma crystallisation history. The relatively stable and blue-dominant CL of zoned phenocrysts is principally related to variations in the Ti concentration, which is a function of the crystallisation temperature. The Al/Ti ratio of igneous quartz increases with progressive magma diﬀerentiation, as Ti is more compatible, compared to Al, Li, K, Ge, B, Fe, P during magma evolution. The red-dominant CL of the anhedral groundmass quartz in granite is unstable during electron bombardment and associated with OH- and H2O-bearing lattice defects. Thus, CL properties of quartz are diﬀerent for rocks formed from H2O-poor and H2O-rich melts. Both groundmass and phenocrysts in granites are rich in alteration structures as a result of interaction with deuteric ﬂuids during cooling, whereas phenocrysts in extrusive rocks do not usually contain such structures. The combined study of trace elements along with the analysis of quartz textures and melt inclusion inventories may reveal detailed PTX-paths of granite magmas. This study shows that quartz is a sensitive indicator for physico-chemical changes during the evolution of silicarich magmas. Common growth textures show a wide variety in quartz phenocrysts in rhyolites and some granites. This paper presents a classiﬁcation of textures, which formed as a result of heterogeneous intra-granular lattice defects and impurities. The alternation of growth and resorption microtextures reﬂects stepwise adiabatic and non-adiabatic magma ascent, temporary storage of magma in reservoirs and mixing with more maﬁc, hotter magma. The anhedral groundmass quartz overgrowing early-magmatic phenocrysts in granites is free of growth zoning. KEY WORDS: adiabatic, cathodoluminescence, crystal resorption, growth textures, lateHercynian magmatism, quartz Quartz provides a detailed chronicle of physicochemical changes in granitic melts. Quartz phenocrysts normally grow in diﬀerent magma batches, which are repeatedly recharged. The solidiﬁed granites and rhyolites therefore comprise mixtures of diﬀerent phenocryst populations embedded in a microcrystalline to coarse-grained groundmass. Feldspars in granites and rhyolites are commonly altered due to the late- to postmagmatic interaction with deuteric and hydrothermal ﬂuids. Quartz is more resistant in this environment, with the consequence that magmatic zoning patterns are commonly preserved. These patterns can be revealed by cathodoluminescence (CL) techniques. Information about the physicochemical changes during the evolution of granitic magma can be obtained by the study of (1) growth patterns in quartz as visualised by cathodoluminescence; (2) trace element concentrations and distribution within quartz crystals; and (3) melt inclusions hosted by quartz: 1. Laemmlein (1930) examined by optical microscope – probably for the ﬁrst time – primary growth structures in quartz phenocrysts in rhyolite from Transbaikalia. It was not until the 1980/90s, however, that growth zoning (Schneider 1993;

2009 The Royal Society of Edinburgh. doi:10.1017/S17556909016144

D’Lemos et al. 1997; Watt et al. 1997) and alteration textures (Sprunt & Nur 1979; Behr 1989; Behr & FrentzelBeyme 1989) in igneous quartz were studied in more detail by CL microscopy. During the last decade, CL microscopy has become a routine method to visualise growth and alteration patterns in igneous quartz (Mu¨ller et al. 2000, 2002a, 2003b, 2005; Peppard et al. 2001; Ruﬃni et al. 2002; Wark et al. 2007; Wiebe et al. 2007). CL reveals episodes of quartz crystallisation including crystal nucleation, growth and dissolution, which develop during magma ascent, storage, recharge, and mixing. The eﬀects of late- to postmagmatic quartz alteration also become visible in CL studies. 2. CL colour and intensity are the result of intrinsic lattice defects (e.g., oxygen and silicon vacancies, broken bonds) as well as trace elements in the crystal structure (e.g., Sprunt 1981; Ramseyer et al. 1988; Perny et al. 1992; Stevens Kalceﬀ et al. 2000; Go¨tze et al. 2001, 2004, 2005). However, the physical background of quartz luminescence has not been fully understood. The concentration of trace elements in quartz is controlled by their abundance in the melt, the partitioning between diﬀerent phases, and the thermo-

186

AXEL MÜLLER ET AL.

Figure 1 The Hercynian Europe according Rajpoot (1991) with sample locations: 1=Erzgebirge/Krusˇne´ Hory; 2=Oberpfalz; 3=Odenwald; 4=Cornwall.

dynamic conditions in the system (Mu¨ller et al. 2000, 2002a, 2003b; Jacamon & Larsen 2009; Wark et al. 2007). Several attempts have been made to combine CL and trace element analysis allowing the quantitative measurement of the Al, Ti, K, and Fe distribution of quartz in relation to CL textures (Watt et al. 1997; Mu¨ller et al. 2000, 2002a, 2003b; Ruﬃni et al. 2002; Wark et al. 2007; Wiebe et al. 2007). In other studies the total content of Al, Ti, Ge, and Li of igneous quartz has been used to discriminate between magmas of diﬀerent origin, or in order to better understand the fractionation process (Lyakhovich 1972; Suttner & Leininger 1972; Schro¨n et al. 1988; Gurbanov et al. 1999; Larsen et al. 2000, 2004; Go¨tze et al. 2004; Jacamon & Larsen 2009). Wark & Watson (2006) established the Ti-in-quartz geothermometer, which allows the calculation of the magma temperature at diﬀerent stages of quartz growth. The incorporation of Al into the quartz lattice is probably controlled by the aluminium saturation index of the magma (Jacamon & Larsen 2009). 3. Igneous quartz, in particular phenocrystic quartz, contains melt inclusions. Melt inclusions comprise a unique record of melt and volatile contents prior to magma emplacement. They record the evolving melt composition and can be used for magma pressure estimates in relation to quartz growth zones. In their study on the Bishop Tuﬀ (eastern California, USA) Peppard et al. (2001) related diﬀerent melt inclusions to growth zones in quartz phenocrysts. Wark et al. (2007) continued their work and applied the Ti-in-quartz thermometer to explain variations of the volatile content in melt

inclusions. They showed that quartz phenocryst rims containing CO2-rich melt inclusions document an w100(C increase of magma temperature shortly prior to eruption. The aim of the present study is to reveal the magmatogenetic signiﬁcance of CL properties and textures in relation to the trace element distribution of igneous quartz in late-Hercynian (Upper Carboniferous to Lower Permian) felsic plutonic and volcanic rocks. Furthermore, the paper compiles the results of a number of earlier studies on growth structures and the microchemistry of igneous quartz (Mu¨ller 2000; Mu¨ller et al. 2000, 2002a, 2003b, 2005, 2006a). Samples were collected from the Erzgebirge, Oberpfalz, and Odenwald (Germany), and from Cornwall (SW England; Fig. 1). As a reference, some intermediate to felsic plutonic rocks and associated volcanic rocks of diﬀerent age and tectonic settings have been included. These rocks (1) are well studied in respect to age, chemical evolution and tectonic setting; (2) cover a wide range of chemical variability from dioritic/dacitic to highly evolved granitic/rhyolitic compositions; (3) include both plutonic and associated volcanic rocks; (4) show well-developed quartz textures (Mu¨ller et al. 2000, 2002a, 2003b, 2005), and (5) have melt inclusion data available (Thomas 1992, 1994; Mu¨ller et al. 2006b). This study combines several visualising and micro-analytical methods, i.e. scanning electron microscope cathodoluminescence (SEM-CL) and optical microscope CL (OM-CL) combined with Fourier-Transform infrared (FTIR) spectroscopy, and electron probe microanalysis (EPMA). FTIR analysis

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES

gives information about the atomic conﬁguration and distribution of hydrogen and structurally bond water in quartz (e.g., Kats 1962; Bambauer et al. 1963; Aines & Rossmann 1984; Kronenberg et al. 1986; Rovetta et al. 1989; Bahadur 1994). The micro-distribution of the trace elements Al, Ti, K and Fe in quartz is determined by EPMA with a spatial resolution down to 5 m. This combination of analytical techniques allows a crystal-chemical characterisation of igneous quartz and therewith a better understanding of large-scale magmatic processes.

1. Methods 1.1. Optical microscope cathodoluminescence (OM-CL) CL spectra and colour images were obtained using a hotcathode luminescence microscope HC3–LM (Neuser et al. 1995). The electron gun operates at a voltage of 14 kV under high vacuum (105-mbar) and a ﬁlament current of 0·18 mA resulting in a current density of ca. 10 mA/mm2 at the sample surface. Photographic documentation was carried out with a NIKON Microﬂex UFX-II system equipped with a NIKON FX-35A reﬂex camera. The spectral response of the luminescence was recorded with a triple-grating (100 lines/mm, 1200 lines/mm, and 1800 lines/ mm) spectrograph TRIAX 320 equipped with a LN2-cooled CCD-detector. The 100 lines/mm grating was used to acquire emission spectra in the range of 400–900 nm (3·1–1·4 eV); the 1200 lines/mm grating corresponds to a spectral range of 70 nm. The CL-emission was collected for an area of w0·5 mm2 using a 20/0·40 objective. The integration time for spectrum acquisition was 20 seconds for the 100 lines/mm grating and 30 seconds for the 1200 lines/mm grating. The spectra were corrected for the total detector response. The CL emission spectra are presented as the sum total of Gaussian curves for the diﬀerent emission lines. These spectra have been deconvoluted using the procedure of Mu¨ller et al. (2002b). The time-dependent variations of the 1·96 and 2·79 eV emissionline intensities during electron radiation were recorded with a f/3·4 Grating Monochromator at a speed of 10 mm/min.

187

1.3. Electron probe microanalysis (EPMA) Trace element abundances of Al, K, Ti, and Fe in quartz were performed with a JEOL 8900 RL electron microprobe with ﬁve wavelength dispersive detectors at the Geowissenschaftliches Zentrum Go¨ttingen. Synthetic Al2O3 (52·9 wt% Al), orthoclase from Lucerne, Switzerland (12·2 wt.% K), synthetic TiO2 (59·9 wt.% Ti), and haematite from Rio Marina, Elba (69·9 wt.% Fe) were used for standards. A beam current of 80 nA, an accelerating voltage of 20 kV, diameter of 5 m, and counting times of 15 seconds for Si, and of 300 seconds for Al, Ti, K, and Fe were used. Raw analyses were converted into concentrations using the phi–rho–Z matrix correction method of Armstrong (1995). Analytical errors were calculated from the counting statistics of peak and background signals, following the Gauss law of error propagation. At low element concentrations the background forms the main part of the total signal. On the other hand, the background signal is nearly constant for a given quartz matrix and the absolute error based on counting statistic is thus nearly constant. Limits of detection by single measurement average (3 ; n=36) are 13 ppm for Al, 13 for Ti, 10 for K, and 15 for Fe.

1.4. FTIR spectroscopy Infrared analysis was performed at the GeoForschungsZentrum Potsdam using a Bruker IFS 66v FTIR-spectrometer coupled with an IR-microscope (Hyperion 1000). The measurements were performed in situ at 190(C using a Linkam FTIR600 heating/freezing stage. The spectra were acquired from 4500 to 2500 cm 1 with a resolution of 2 cm 1. Analyses were carried out on doubly-polished wafers of quartz crystals (diameter about 3 mm) with a thickness of 50–80 m. The beam diameter on the sample ranged from 30 m to 40 m and allows the detection of variations within individual quartz crystals. Four to eight points along a line in one crystal could be measured. Thin quartz wafers were used to minimise the risk of encountering ﬂuid inclusions, which are common in quartz from subvolcanic and plutonic rocks. Prior to the analysis the wafers were cleaned with acetone and dried for 10 h at 110(C. The spectra were corrected for the wafer thickness.

1.2. Scanning electron microscope cathodoluminescence (SEM-CL)

2. Cathodoluminescence of igneous quartz

Two scanning electron microscopes equipped with diﬀerent CL detectors, the Cambridge Instruments 250-MK3 with a S20-Extended photomultiplier and the JEOL JXA 8900 with a CLD40 R712 photomultiplier, were used for the study of microtextures in quartz. The detectable wavelength for both photomultipliers ranges from 380 to 850 nm (3·26–1·46 eV). SEM-CL allows a better spatial resolution down to 1 m2 and the capability of combining CL with back scattered electron (BSE) imaging and electron microprobe analysis (EPMA). The possibility of increasing the power density over small sample areas is useful for samples with low CL intensity. A disadvantage is the monochromatic (grey scale) image. Weakly luminescing quartz corresponds commonly to red to reddish brown CL and bright quartz to blue to violet colours, with the present instrument. Images were collected from the JEOL system using a beam voltage of 30 keV, a ﬁlament current of 200 nA, slow beam scan rates of 20 seconds at processing resolution of 1024860 pixels and 256 grey levels. The voltage and sample current for the Cambridge Instruments 250-MK3 was 20 keV and 5–15 nA, respectively. The documentation of the CL images at the 250-MK3 were carried out with a photo camera with Agfapan APX 25 ﬁlms and by using slow beam scan rates of 250 seconds. The contrast of the images was improved by using the software PhotoShop for Windows.

The cathodoluminescence of igneous quartz commonly shows shades of blue, violet, red and red-brown. These colours mainly result from emissions in the red (1·7–2·2 eV) and blue (2·4–3·1 eV; Fig. 2). Contrary to most other luminescing minerals, the defect structure inventory of quartz, and therewith the cathodoluminescence, is inﬂuenced by the electron beam. The ratio of the red and blue emission ranges changes with the irradiation time, whereby the blue emission normally decreases and the red emission increases (Fig. 2). As a consequence, the initial CL signal is diﬃcult to detect. The radiation damage of the upper ca. 5 m of the quartz sample surface means that the luminescence is non-reproducible. High-resolution spectral analysis shows that both red and blue emissions are composed of several emission lines. Three emission bands are distinguished in the red emission range (1·73 eV, 1·84 eV, and 1·96 eV), and ﬁve bands in the blue emission range (2·47 eV, 2·58 eV, 2·68 eV, 2·79 eV and 2·96 eV). A broad yellow band has been identiﬁed at 2·15 eV (Table 1; Fig. 2). The individual CL emission bands can be localised and quantiﬁed from best-ﬁt Gaussian curves by working out about ca. 100 spectra (Mu¨ller 2000). The CL band positions are reported also in the literature (e.g., Remond et al. 1992; Gorton et al. 1996; Stevens Kalceﬀ et al. 2000 and references therein; Table 1).

188

AXEL MÜLLER ET AL.

Figure 2 CL spectra of igneous quartz recorded with the 100 lines/mm grating after 30 seconds, two and seven minutes of electron bombardment. The seven-minute-spectra were resolved by best ﬁtting with Gaussian curves.

The blue 2·68 eV and 2·58 eV bands have a constant band area ratio of 1:0·340·08. Similar observations were made for the red 1·96 eV and 1·84 eV bands, which show a ratio of 1:0·510·17. Predominant emission bands of zoned quartz phenocrysts are the 1·96 eV, 2·58 eV, 2·68 eV, 2·79 eV and 2·96 eV band. Exceptions are the phenocrysts from the Eibenstock and Aue granite from the Western Erzgebirge, which show a characteristic 2·15 eV emission. Some growth zones in the phenocrysts show a marked 1·73 eV emission. The growth zones within quartz phenocrysts correspond to variable intensity of the blue emission, whereas the red emissions are approximately constant. The blue emission (2·6–3·1 eV) shows high intensity at initial electron bombardment, but a part of the blue emission lines mostly centred at w2·8 eV (2·58 eV, 2·68 eV, 2·79 eV and 2·96 eV), drop by 1/2 to 1/3 after a few seconds. However, the 2·79 eV emission decreases much more than the 2·58 eV, 2·68 eV and 2·96 eV emissions (Mu¨ller 2000). After 30 seconds

to 100 seconds of electron bombardment the CL is stabilised, pointing at a constant number of luminescence centres at the experimental conditions. The applied experimental conditions resulted in a beam current density of w10 mA/mm2 at the sample surface and temperature increase of ca. 110(C during 100 seconds of initial bombardment. The red emissions (1·75– 2·2 eV) show lowest intensity at initial electron bombardment followed by a steep parabolic increase during the ﬁrst minute of electron radiation followed by a slight increase until a stable luminescence is reached. Most of the energy of the electron beam is transformed into heat in the sample (e.g., Remond et al. 1992). By using the present equipment, the temperature of the sample may increase by up to 140(C after 10 minutes of electron beam irradiation (Mu¨ller 2000). Therefore, thermoluminescence bands may impose CL spectra. Characteristic thermoluminescence bands are at 2·95–2·85 eV, 2·21–2·14 eV and 2·00–1·98 eV (e.g., Jani et al. 1983; Rink et al. 1993; Yang et al. 1994). The intensity

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES

189

Figure 2 (Continued).

maxima of the 2·21–2·14 eV and 2·00–1·98 eV bands lie in the temperature range of the CL measurements. However, the marked diﬀerent CL for diﬀerent quartz types in the same temperature range demonstrates that the eﬀect of temperature is relatively small (Mu¨ller 2000). The kinetic law c=c0 e(kt) (c=concentration of the component at time t, c0 =concentration of the component at time t=0, and k=velocity [equilibrium] constant) has been applied in order to quantify the changes of CL during electron beam irradiation (Ramseyer et al. 1988; Picouet 1999); the increase of the red CL can be described by the reverse equation of the kinetic law. In this way, the CL intensity can be approximately quantiﬁed as a function of time. According to Ramseyer et al. (1988) and Picouet (1999) the fastly decaying component and the slowly decaying components in the blue CL-range are calculated as the sum of two kinetic law equations: Ib =Ibs +Ib1 e(t/kb1)+Ib2 e(t/kb2)

(1)

where Ib =intensity of blue CL at the radiation time t; Ibs =intensity of stable blue CL; Ib1 =intensity of the slow decreasing CL component at t=0; kb1 =velocity constant of the slow decreasing CL component; Ib2 =intensity of the fast decreasing CL component at t=0; kb2 =velocity constant of the fast decreasing CL component; t=radiation time. The increase of red emission can be described by the equation: Ir =Irs Ir1 e(t/kr1)Ir2 e(t/kr2)

(2)

where Ir =intensity of red CL at the radiation time t; Irs =intensity of red CL for t/N; Ir1 =intensity of the slow increasing CL component for t/N; kr1 =velocity constant of the slow increasing CL component; Ir2 =intensity of the fast increasing CL component for t/N; kr2 =velocity constant of the fast increasing CL component; t=radiation time. The velocity constants kr1, kr2, kb1, and kb2 depend on the measurement conditions and are similar for the present

190

AXEL MÜLLER ET AL.

Table 1 Detected CL emission bands of igneous quartz (1·4–3·1 eV) and their possible association with trace elements and defect centres according to literature. Band position (eV)

Half width (eV)

1·730·02

0·30·02

1·840·01 1·960·02

0·220·01 0·220·02

2·150·02

0·380·01

2·470·02

0·300·03

2·580·01 2·680·01 2·790·01 2·960·015

0·180·005 0·230·01 0·260·01 0·300·02

Proposed origin

Reference with band position

substitutional Fe3+

Pott & McNicol (1971) – 1·78 eV Kempe et al. (1999) – 1·75 eV nonbridging oxygen hole centre Sigel & Marrone (1981) – 1·85 eV oxygen vacancy Luﬀ & Townsend (1990) – 1·91 eV nonbridging oxygen hole centre with OH-groups as precursors Stevens Kalceﬀ & Phillips (1995) – 1·95 eV E# centre [SiO3]3 associated with substitutional Ge Go¨tze et al. (1999) – 2·1 eV Luﬀ & Townsend (1990) – 2·18 eV extrinsic defect [GeO4/Li + ]0 Stevens Kalceﬀ & Phillips (1995) – 2·46 eV Go¨tze et al. (2005) – 2·45 eV intrinsic defect Gritsenko & Lisitsyn (1985) – 2·58 eV self trapped exciton combined with E# centre [SiO3]3 Stevens Kalceﬀ & Phillips (1995) – 2·68 eV self trapped exciton Itoh et al. (1990) – 2·8 eV intrinsic defect associated with substitutional Ti Stevens Kalceﬀ & Phillips (1995) – 2·93 eV Mu¨ller et al. (2002b) – 2·96 eV

Figure 3 Plot of the ﬁtted parameters Ir2 and Ib2 representing fast changing CL components of the red CL (1·96 eV) and blue CL (2·79 eV), respectively. Granitic groundmass quartz exhibit the most unstable CL characterised by a strong increase of the red CL.

samples. Mu¨ller (2000) showed that for diﬀerent quartz types the decay of the luminescence centres causing blue CL is about two times faster than that of those causing red CL. The emission intensities (I) correspond to the concentration of luminescence centres in the interaction volume of the electron beam. The parameters Irs, Ir1, Ir2, Ibs, Ib1, and Ib2 are the intensity portions of all emissions, whereby Ir1, Ir2, Ib1, and Ib2 are intensity portions of the changeable CL. In granites, zoned quartz phenocrysts have low Ir2 and Ib2, which is indicative of more stable defect structures compared to the groundmass quartz (Fig. 3). The latter quartz exhibits widely varying intensities reﬂecting a high concentration of unstable defect structures. The high Ir1 and Ir2, which is characteristic for granitic groundmass quartz, demonstrates the predominance of unstable red CL.

3. Intra-granular textures of igneous quartz 3.1. Primary growth textures The term ‘primary growth textures’ is used here for intragranular growth patterns which develop during crystal growth in the magma; secondary structures are the result of alteration. Primary growth textures may develop in (1) early-magmatic quartz phenocrysts, (2) late snowball quartz in highly evolved

albite granites (e.g., Mu¨ller & Seltmann 1999), and (3) comb quartz in layered granitic rocks (e.g, Mu¨ller et al. 2002a). 3.1.1. Early-magmatic quartz phenocrysts. Quartz phenocrysts have been described worldwide from rhyolites, rhyodacites and some dacites with R63 wt.% SiO2. In dacites or other more maﬁc volcanic rocks, quartz phenocrysts are commonly rounded and sometimes show coronas of hornblende, pyroxene and/or An-rich plagioclase (oscelli texture), indicating disequilibrium with the melt (e.g., Mashima 2004). They must have been inherited from a more felsic magma prior to emplacement or extrusion. Phenocrysts in volcanic rocks can be easily recognised due to the grain size contrast between them and the micro- to ﬁne-crystalline groundmass. However, groundmass minerals in plutonic rocks are coarser grained and may overgrow phenocrysts. The early magmatic crystals may be not phenocrysts in the strict sense because they may not have shared common histories and crystallised from a melt in which they are now hosted, but rather represent a melt laced with a crystal cargo that has been inherited from melts that existed at diﬀerent places and times in the magma system (Davidson et al. 2007). Neither are they strictly xenocrysts, as they are grown and recycled from closely related progenitor magmas rather than accidentally incorporated from unrelated wallrocks. W. E. Hildreth (pers. comm. in Davidson et al. 2007) suggested the term ‘antecryst’ to denote phases that originate in the magma system, but are not true phenocrysts (cf. Charlier et al. 2005). Davidson et al. (2007) proposed that magmatic rocks can be represented by mixtures of melts, recycled antecrysts, and true phenocrysts. However, in this present study, the dominant early magmatic crystal populations of grantic rock are discussed and, therefore, they are considered as true phenocrysts. In plutonic rocks phenocrysts are often diﬃcult to identify, but can be identiﬁed by their contrasting CL. The majority of the late-Hercynian granites with R65 wt.% SiO2 contain quartz phenocrysts (Mu¨ller et al. 2000, 2002a, 2003b, 2005, 2006a; Table 2). In some granites the phenocrysts comprise a substantial part of the rock up to 40 vol.% (e.g., Mu¨ller et al. 2000), whereas in other granites phenocrysts are rare or even absent. A number of late-Hercynian granites with R63 wt.% SiO2 contain exclusively anhedral groundmass quartz without zoning, e.g. the Flossenbu¨rg and Leuchtenberg granite from the Oberpfalz. Groundmass quartz may show euhedral crystal cores with brighter CL, e.g. Rozvadov granite, Oberpfalz. Migmatitic granites, chemically highly evolved albite granites

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES

191

Figure 4 Schematic illustration of growth textures observed in early-magmatic quartz phenocrysts of lateHercynian felsic igneous rocks. 1a–d=step zones; 2=oscillatory zone; 3=skeletal growth; 4=wavy surface; 5a=melt inclusion with negative crystal shape; 5b=ovoid melt inclusion; 6a=lobate growth embayment; 6b=closed growth embayment; 7=resorption surface; 8=roundmass quartz; 9=secondary structures.

and ﬂuid-enriched aplites and miarolitic granites are generally free of early-magmatic quartz phenocrysts. The development of zoning in phenocrysts requires that the quartz grew in an evolving magma reservoir. Microscopic growth zone boundaries represent relic crystal-melt interfaces, which develop by ﬂuctuations of growth and diﬀusion rates. The zoning pattern of the phenocrysts is normally more complex in granites and rhyolites with high SiO2 content (>70 wt.%). This can be explained by the early SiO2 saturation with the consequence of a longer period of quartz precipitation prior to ﬁnal magma emplacement or extrusion. Phenocrysts in granites and rhyolites show similar growth patterns. In some cases, quartz phenocryst populations in granites are similar to populations in the associated rhyolites (e.g., Schellerhau granite and Teplice rhyolite; Mu¨ller et al. 2005). However, the zoning of granitic phenocrysts is normally less well developed. The zoning may have been quenched or overprinted during slow magma cooling and be destroyed by the interaction with deuteric ﬂuids. Growth patterns in quartz of granites and rhyolites have been described in Mu¨ller et al. (2000, 2003b, 2005). The present authors distinguish ﬁne-scale (2–20 m) oscillatory zoning, large-scale step zoning (compositional zoning), skeletal growth, wavy and resorption surfaces and growth embayments (Fig. 4).

The SEM-CL images of Figures 5 and 6 display characteristic growth textures in igneous quartz from late-Hercynian granites and rhyolites from Central Europe. The images represent a selection of about 1400 pictures taken from more than 260 thin sections. The examples in Figures 5 and 6 represent common features in these rocks. Each igneous system has its own distinctive ﬁngerprint as recorded in the zoning pattern (Table 2). Textures indicate that epitaxy must be important for quartz nucleation. Quartz nucleates on tiny grains of feldspar, mica, accessory minerals, or on immiscible melt droplets or bubbles occasionally preserved as melt inclusions (Fig. 5b, c). Sometimes two or more quartz crystals grow around one particle. Particularly in granites, these crystals may coalesce to become one crystal or they grow separately forming clusters of two to ﬁve crystals (Fig. 5b). Phenocrysts nucleate commonly as dihexagonal -quartz. Rhombohedral -quartz has been described by Flick (1987) from rhyolites from the Odenwald. Phenocrysts from the Odenwald show skeletal overgrowths in their cores (Fig. 5c). Skeletal growth reﬂects a kinematicallydriven, non-equilibrium growth phase (e.g., Kirkpatrick 1981; Fenn 1986; Fowler 1990) which is in accordance with the undercooling experiments of Fenn (1977) and Swanson (1977), who showed that growth is controlled by diﬀusion during rapid crystallisation.

192

AXEL MÜLLER ET AL.

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES

193

Table 2 Characteristic features of quartz phenocrysts in late-Hercynian igneous rocks. Late-Hercynian igneous province

Investigated igneous rocks

Rock type

SiO2 wt.%

Characteristic growth structures of phenocrysts

Eastern Erzgebirge

Eastern Erzgebirge volcano-plutonic complex

granodiorite, monzogranite, rhyodacite, rhyolite

69–75

Western Erzgebirge

Eibenstock Nejdek pluton

monzogranite

68–78

phenocryst with complex, contrast-rich zoning, low CL rounded crystal cores, Ti-rich margin (reverse zoning), zoning with -quartz habit (Figs 5a, d, 6e, g) phenocrysts with complex, contrast- poor zoning, zoning with -quartz habit

Oberpfalz

Rozvadov, Leuchtenberg, Flossenbu¨rg, and Falkenberg granites

tonalite, diorite, redwitzite, granodiorite, monzogranite

52–76

no phenocryst or phenocrysts with simple, contrast-poor zoning, zoning with -quartz habit

Odenwald

Weinheim, Dossenheim

rhyolite

71–74

phenocryst with complex, contrast- rich zoning, skeletal growth, zoning with -quartz habit (Fig. 5c)

Cornwall

Land’s End granites

granite

66–77

phenocrysts with contrast-poor zoning, zoning with -quartz habit

Correlative zoning patterns of quartz phenocrysts from the same magma batch indicate that the crystals of one population nucleated simultaneously and continued growing together (Peppard et al. 2001; Mu¨ller et al. 2000, 2005). Some rocks contain two or more phenocryst populations, whereby one population dominates. Each population represents one nucleation event. In this way the temporal crystallisation history of a magma batch can be reconstructed (Mu¨ller et al. 2000). Due to multiple magma recharge, quartz populations with a diﬀerent history can be mingled (Mu¨ller et al. 2005). This is known from plagioclase in maﬁc and intermediate rocks (e.g., Davidson et al. 2001). The ﬁrst growth stage of quartz phenocrysts may be interrupted by resorption, resulting in rounding of the crystal nuclei. The resorption surface truncates pre-existing growth zones and results in the rounding of crystals. This process is recorded by rounded crystal cores with weak redbrown CL as commonly found for the Eastern Erzgebirge granites and rhyolites (Fig. 5a, d). Major resorption surfaces are typically overgrown by bright blue luminescing quartz (Mu¨ller et al. 2003b). Sometimes the

Schemes of growth structures of dominant phenocryst populations (oscillatory zoning is not shown)

blue luminescing overgrowth is very thin (<10 m) and may show a dendritic or wavy structure (Fig. 6a). These overgrowths may be enriched in ovoid melt inclusions (Fig. 5g). Fine-scale (1–5 m) oscillatory growth zones with low amplitude in CL are sometimes separated by zones recording resorption or skeletal overgrowth, accompanied by a major change of the CL properties. These events with high-amplitude CL superimpose on the oscillatory zoning. Growth stages of low amplitude CL bordered by resorption or skeletal growth are considered step zones (Fig. 5b; Alle`gre et al. 1981). Analogous distinction between low-amplitude (An-content) ﬁne-scale zoning and superimposed high-amplitude large-scale zoning has been made for plagioclase phenocrysts (Pearce & Kolisnik 1990). The quartz phenocryst of Figure 5b shows two step zones. The deﬁnition of step zones becomes more complicated for phenocrysts with complex growth patterns. For example, six step zones can be distinguished in the phenocryst shown in Figure 5c. 3.1.2. Late-magmatic snowball and comb quartz. Highly evolved topaz-bearing albite granite contains typically

Figure 5 SEM-CL images of early-magmatic quartz phenocryst from late-Hercynian felsic igneous rocks illustrating the variability of primary growth textures: (a) Zoned quartz phenocryst embedded in groundmass quartz. Schellerhau granite SG2; (b) Cluster of three zoned phenocrysts of the Gattersburg rhyolite (NW Saxony, Germany). The left crystal germinated at melt droplet which is now preserved as melt inclusion (white arrow). The growth pattern is characterised by two step zones with subordinate oscillatory zoning. The relatively homogenous pattern indicates almost undisturbed crystal growth. The phenocrysts show several thin healed contraction cracks (black) caused by shock cooling during eruption; (c) Rhyolitic phenocryst with -quartz habit with complex growth pattern (Dossenheim, Odenwald). The bright crystal core, which germinates at a melt/volatile droplet (black arrow), shows skeletal growth (white arrow); (d) Rhyolitic phenocryst (Hengstberg, NW Saxony) with dull luminescing, resorbed crystal core (black arrow) with bright overgrowth (white arrow). The euhedral phenocryst is overgrown by almost non-luminescing groundmass quartz (black); (e) Fracture of zoned phenocryst with weak luminescing core (Teplice rhyolite, Erzgebirge). Circles correspond to FTIR absorption sampling spots; (f) Rhyolitic phenocryst (Fichtelgebirge) with a major resorption surface (white arrow). Subsequent wavy growth show bright luminescence; (g) Rhyolitic quartz phenocryst (drill core KB1 Wittichen 263–264 m, Black Forest). The resorption event (white arrow) is followed by irregular growth which result in the entrapment of numerous large melt inclusions (MI); (h) Phenocryst of the Beucha granite porphyry (NW Saxony). The resorbed cluster of zoned phenocrysts (black arrow) is embedded in bright luminescing anhedral groundmass quartz.

194

AXEL MÜLLER ET AL.

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES

195

acteristic for early-magmatic phenocrysts, have not been observed in the snowball quartz. Comb quartz occurring typically in granitic line rock nucleates at intra-granite contacts (Breiter et al. 2005). It has similar CL properties and structural features as the snowball quartz, with the diﬀerence that it does not form double-ended crystals (ﬁg. 3c in Mu¨ller et al. 2002a).

3.2. Secondary CL structures

Figure 7 Growth structures of late-magmatic snowball quartz with zonal enclosed albite laths (white). Modiﬁed from Mu¨ller et al. 2002a.

so-called snowball quartz (Beus et al. 1962; Kovalenko 1977; Sonyushkin et al. 1991; Poutiainen & Scherbakova 1998; Mu¨ller & Seltmann 1999; Mu¨ller et al. 2002a; Fig. 6b, c). These rocks occur in the apical part of sub-volcanic granite complexes and are commonly associated with rare-metal mineralisation. Snowball quartz has a dull red-brown to bright red/orange CL, in contrast to early-magmatic quartz phenocrysts, which have predominantly bluish to violet CL. The crystals show gradual transitions into the ramiﬁed dendritic crystal margin and into the matrix quartz without changes of the CL properties. Older quartz generations, which may be present in the albite granites are cannibalised by snowball quartz as shown in Figure 6c. Inclusions of groundmass minerals, such as corroded K-feldspar, mica, zircon, apatite and particularly albite are enriched in the growth zones. Fluid and melt inclusions are abundant throughout the snowball quartz. Lath-shaped albite crystals wrap around the snowball quartz, indicating that the quartz pushed aside the albite laths during growth in a dense crystal mush. The snowball quartz shows oscillatory growth zoning with low CL contrasts. The zone boundaries are characterised by planar growth zones with -quartz crystal habit. Resorption surfaces, which are char-

Secondary structures in the CL observation mode are principally formed by ﬁve processes: (1) micro-cataclasis followed by healing; (2) stress-induced and ﬂuid-driven quartz dissolution and precipitation; (3) stress-induced lattice reorganisation (puriﬁcation); (4) impurity element diﬀusion; and (5) -radiation. In a cooling magma late- to post-magmatic ﬂuids may result in small-scale quartz dissolution–precipitation along grain boundaries and micro-fractures. Dense healed fractures connecting low-luminescing domains around ﬂuid inclusions are widespread in granitic rocks (Sprunt & Nur 1979; Behr & Frentzel-Beyme 1989; Valley & Graham 1996; D’Lemos et al. 1997; Van den Kerkhof & Hein 2001, Van den Kerkhof et al. 2001, 2004; Mu¨ller et al. 2000, 2002b; Rusk & Reed 2002; Fig. 6e, i, j). Micro-fracturing can be attributed to internal stresses at a grain scale, resulting from the thermal contraction of quartz relative to feldspar in the cooling granite (Vollbrecht et al. 1991, 1994). The / -quartz transformation which causes an anisotropic contraction of 0·86 vol.% vertical to the c-axis and 1·3 vol.% parallel to the c-axis (e.g., Blankenburg et al. 1994), imposes additional stress. However, the / -quartz transformation is not the dominant process causing the fracture network because these structures are trans-granular, meaning they are independent from the crystallographic orientation of the quartz crystals, and they are absent in phenocrysts from extrusive rocks (Fig. 6e, g). Mu¨ller (2000), Mu¨ller et al. (2002) and Van den Kerkhof et al. (2004) demonstrated the systematic reduction of Al, Ti and K in secondary quartz compared to the host quartz. However, the CL intensity of secondary quartz may increase after several minutes of electron beam exposure using high beam power densities >104 W/cm2 (Fig. 6i, j) resulting in contrast reversal. Partially healed radial and concentric contraction cracks caused by shock cooling are common in quartz phenocrysts from volcanic rocks (Fig. 6a). Groundmass quartz in granites, which is free of primary growth zoning, commonly exhibits non-luminescing 1–5 m small micropores, which become visible as dark spots in SEM-CL after several minutes of electron bombardment (Van den Kerkhof & Grantham 1999, Mu¨ller 2000, Van den Kerkhof & Hein 2001). Figure 6d shows such spots in quartz from the Niederbobritzsch granite, Eastern Erzgebirge. Igneous quartz sometimes shows halos of pinkish/yellowish white luminescing quartz around zircon inclusions. These textures are caused by -radiation damage, i.e. metamictisation

Figure 6 SEM-CL images of quartz in igneous rocks illustrating the variability of secondary structures in early-magmatic quartz phenocryst and textures of late-magmatic groundmass and snowball quartz: (a) Rounded (resorbed) quartz phenocryst of the Scho¨nfeld rhyodacite (Eastern Erzgebirge). The internal growth zoning is blurred. The crystal is overgrown by a very thin layer of bright luminescing (white) dendritic groundmass quartz; (b) Snowball quartz of the Schellerhau granite SG3, Erzgebirge, with numerous albite inclusions; (c) Snowball quartz (sq) cannibalising (white arrow) groundmass quartz (gq) in the Podlesı´ Dyke granite, Erzgebirge. zw=zinnwaldite; (d )Weak luminescent spot-like structures in quartz (Niederbobritzsch granite, Erzgebirge); (e) Zoned quartz phenocryst in a subvolcanic rock (Teplice rhyolite TR3c, Erzgebirge). In contrast to (g), the quartz contains a dense network of secondary structures (black); (f ) BSE image of the surface of (e) with high porosity (up to 2 vol.%); (g) Zoned quartz phenocryst from an eﬀusive rock (Teplice rhyolite TR2b, Erzgebirge). (h) Back-scattered electron (BSE) image of the quartz surface of (g) showing very low porosity (<0·2 vol.%); (I) Network of almost non-luminescing domains and healed cracks in the Schellerhau granite (Erzgebirge) at initial electron bombardment (j) The area shown in (i) after 10 minutes focused electron bombardment (20 keV, 15 nA). The luminescence of the secondary quartz had been increased drastically.

196

AXEL MÜLLER ET AL.

trations of 20 to 50 ppm (Mu¨ller et al. 2002a, 2003a, b). An exponential Al/Ti trend can be recognised (linear trend on the log–log plot of Fig. 8), which is however not continuous. The high-Al tails of the data ﬁelds of the Scho¨nfeld basal rhyolite, the Teplice rhyolite (Mu¨ller et al. 2002a) and the rhyolite from Bonne Nuit Bay (Watt et al. 1997) are caused by a few out-of-range values (Fig. 8a), which are the result of submicroscopic inclusions (see section 6.1). Taking this into account, Al concentrations in quartz phenocrysts are relatively constant for each volcanic facies, compared to Ti. Al is most variable in evolved granites of the Western Erzgebirge (Podlesı´ Dyke granite, Aue granite, Slavkov microgranites). Concentrations of Li, K, Ge, B, Fe, and P in igneous quartz are in the range of 1 to 50 ppm (Schro¨n et al. 1988; Go¨tze et al. 2004; Larsen et al. 2004; Jacamon & Larsen 2009). All these elements are incompatible and correlate positively with Al, implying that their concentrations increase with progressive diﬀerentiation of the magma. Li in quartz may be buﬀered by Li-bearing micas, which may coexist with quartz at higher degrees of fractionation (Mu¨ller et al. 2008), whereas B tends to separate into the magmatic vapour phase (e.g., Pichavant 1981). Secondary quartz is typically depleted in Al, Ti, and K (Van den Kerkhof & Mu¨ller 1999; Van den Kerkhof et al. 2004; Mu¨ller et al. 2008).

5. FTIR absorption of igneous quartz

Figure 8 Compilation of Al concentrations and Al/Ti ratios in quartz from (a) dacites and rhyolites and (b) granites. Data were obtained by EPMA and SIMS: (1) Watt et al. 1997; (2) Mu¨ller et al. 2003a; (3) Mu¨ller et al. 2005; (4) Mu¨ller et al. 2000; (5) Mu¨ller et al. 2002a; (6) Mu¨ller et al. 2003b.

of the quartz crystal structure (e.g., Botis et al. 2005). The radius of halos is typically about 40 m. Each halo is subdivided into a brighter inner zone (w25 m) and an outer zone (w15 m). The radius of w40 m corresponds to the interaction radius of -particles in quartz (Owen 1988).

4. Trace elements in igneous quartz Al and Ti are the most abundant trace elements in igneous quartz (Mu¨ller et al. 2000, 2002a, 2003a, b). The contrasting behaviour of these two elements during magma diﬀerentiation make them suitable for discriminating igneous units. Figure 8 shows Al/Ti vs. Ti discrimination diagrams for igneous quartz from dacites, rhyodacites and rhyolites (Fig. 8a) and from granites (Fig. 8b). Quartz in the most primitive rocks (dacites) has the lowest Al and highest Ti (lowest Al/Ti; Fig. 8a). Ti in quartz decreases and Al increases with progressive diﬀerentiation. Ti shows maximum concentrations of up to 130 ppm in the bright blue luminescing growth zones in early-magmatic quartz phenocrysts, whereas the Ti concentration in latemagmatic groundmass quartz in granites shows lower concen-

The abundance of hydrogen-related defects in igneous quartz has been measured by FTIR spectroscopy. Unpolarised spectra were collected along traverses across quartz crystals (Fig. 9). All spectra show distinct absorption at 3365 cm 1 and a minor absorption at 3305 cm 1. Both bands are attributed to hydrogen-compensated Al defects (e.g., Kats 1962). Intense Al absorption was found for the phenocryst cores of the Teplice rhyolite, whereas Al-absorption is less pronounced in quartz from the Scho¨nfeld rhyodacite and from the A r land rapakivi granite. Quartz with blue CL has also relatively low total Al concentrations, as determined by EPMA (80– 130 ppm; see Mu¨ller et al. 2005). On the other hand, the snowball quartz (Podlesı´ Dyke granite) with highest total Al concentrations (Fig. 8b) shows low Al–H-related absorption. Quartz from the Podlesı´ Dyke granite (groundmass quartz), Schellerhau granite and Teplice rhyolite (phase TR1; Mu¨ller et al. 2005) shows a distinct absorption band at 3474 cm 1, which is caused by a proton-compensated Al defect, perturbed by Li (Kats 1962). The groundmass quartz of the Podlesı´ Dyke and Schellerhau granites is characterised by broad absorption bands at 3110 and 3220 cm 1. Both samples show high absorption intensities of both bands, which are assigned to the symmetrical stretching vibrations of frozen molecular water (Aines & Rossman 1984). Low absorption of frozen water exhibits the red luminescing phenocryst core of the Teplice rhyolite (phase TR2b; Mu¨ller et al. 2005). Blue luminescing quartz and snowball quartz do not show water-related absorption. These bands cannot be attributed to condensated ice in the cooling stage. The bands have been only recognised for secondary quartz and some red luminescing growth zones. Similar absorption bands as frozen molecular water may point at water-bearing micropores; the analysed quartz volume is free of visible ﬂuid inclusions. Sharp absorption bands at 3365 cm 1, 3305 cm 1 and 3474 cm 1 and broad absorption bands 3110 cm 1 and 3220 cm 1 as recorded for the cores and the rims of zoned quartz (phenocrysts, snowball quartz) and in unzoned crystals (groundmass quartz) do not show distinct variation. However,

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES

197

Figure 9 FTIR absorption spectra of igneous quartz. Spectra are oﬀset for clarity.

variations of the absorption pattern were found between diﬀerent quartz types in the same rock e.g., between groundmass quartz and snowball quartz (Podlesı´ Dyke granite) and between the primary (igneous) quartz and secondary quartz.

6. Discussion 6.1. CL properties and trace elements of igneous quartz Cathodoluminescence and trace element studies show that hydrogen-related defect structures are abundant in igneous quartz. According to Stevens Kalceﬀ & Phillips (1995) the red

CL emission around 1·96 eV is related to non-bridging oxygen hole centre (NBOHC, an oxygen dangling bond; hSi–O·) with hydrogen as precursor (hSi–OH). Defect structures with hydroxyl are strongly aﬀected by the temperature and ionising radiation: protons and hydroxyl are released and concentrate as molecular water in larger defect structures and micropores (Stenina et al. 1984; Heggie 1992). The increase in the red emission of igneous quartz during electron bombardment is explained by radiolysis of hydroxyl groups in the quartz lattice, which results in the formation of NBOHC. The rate of the red CL increase is moderate for early-magmatic quartz phenocrysts, but very high for quartz in the granitic ground mass.

198

AXEL MÜLLER ET AL.

Small, nearly non-luminescing 1–5 m small spots have been observed in granitic groundmass quartz during electron bombardment at high beam current (e.g., Fig. 6d). Stenina et al. (1984) described similar spots in quartz, which was intensively exposed to electron radiation. By using TEM imaging, these authors identiﬁed these structures as amorphous (non-crystalline) micro-domains, which contain molecular water. The dull-luminescing spots may be explained by diﬀusion of protons, hydroxyl and probably also molecular water migrate (‘jump’) along weak donor-acceptor bonds through the quartz lattice under excitation (Heggie 1992). At defects, hydroxyl and molecular water are trapped, forming water-rich disordered domains. This disorder results in the local quenching of the cathodoluminescence. The elevated concentrations of molecular water in granitic groundmass quartz and secondary quartz as detected by FTIR spectroscopy conﬁrm the existence of water-bearing micropores. The non-luminescing micro-domains and molecular water have not been detected in early-magmatic phenocrysts with dominant blue CL. The relatively homogeneous concentration of structural hydroxyl in the quartz phenocrysts can be explained by the fact that the amount of structural hydroxyl in quartz correlates with substitutional Al. In summary, the structurally bound hydrogen and water is lowest in the early-magmatic phenocrysts and highest in the granitic groundmass. This observation veriﬁes progressively higher water content of the melt during magma evolution. The strong increase of red-CL and the formation of waterrich disordered micro-domains in granitic groundmass quartz reﬂects crystallisation from a cool and water-enriched melt. Consequently, late-magmatic snowball quartz should contain highest concentrations of structural water. However, lowest structural water was found here. This unexpected result is related to very rapid quartz crystallisation in F-, B-, and/ or P-enriched magmas, because hydrogen diﬀusion within the quartz lattice requires time to operate. The low-pressure (subsurface) quartz growth would add to the eﬀect of rapid growth, because the hydroxyl content in quartz is a positive function of pressure (e.g., Mosenfelder 2000). In this speciﬁc environment quartz incorporates only traces of hydroxyl which are below the detection limit of FTIR spectrometry. The ‘ﬂash’ of blue CL commonly observed during the ﬁrst seconds of electron bombardment is caused by an emission band at 3·18–3·26 eV (Alonso et al. 1983; Luﬀ & Townsend 1990; Perny et al. 1992; Gorton et al. 1996) According to these authors the intensity of the emission band correlates well with the Al content and the concentration of paragenetic [AlO4/ M + ]. The band position lies outside the sensitivity range of the spectrometer used in this present study. About 1600 analyses of total Al concentration in quartz do not show a clear correlation with one of the nine emission bands, which were detected for igneous quartz between 1·4 eV and 3·1 eV (Mu¨ller 2000; present paper). This lack of correlation may be attributed to the diﬀerent atomic conﬁgurations of Al in the quartz lattice. The substitutional trivalent Al3+ can be compensated by coupled substitution of a pentavalent ion e.g. P5+ , or by interstitial monovalent ions such as Li + , H + , Na + and K + . Therefore, Al in quartz depends on the availability of the compensating ions in the crystallising melt. The diﬀerent valences of Al in quartz require diﬀerent uptake conditions. The larger part of the total Al in quartz is probably incorporated in sub-microscopic defect clusters (<0·2 m) as discussed by Mu¨ller et al. (2003a). The higher concentration of defect clusters as detected here particularly for quartz with >500 ppm

Al, explains the contradictory low Al-H infrared absorption and the high total Al in the snowball quartz from Podlesı´. The intensities of the detected blue emissions at 2·58 eV, 2·68 eV and 2·79 eV also decrease during electron bombardment and contribute to the initial blue ‘ﬂash’. Correlations between the 2·58 eV, 2·68 eV and 2·79 eV emissions and the total content of Al, Ti, K, Fe could not be found. The 2·68 eV and 2·79 eV emissions are due to the recombination of a so-called self-trapped exciton (Stevens Kalceﬀ & Phillips 1995), whereas the 2·58 eV band cannot be fully explained (e.g., Gritsenko & Lisitsyn 1985). The self-trapped exciton involves an irradiation-induced electron hole pair (oxygen Frenkel pair consisting of an oxygen vacancy and a peroxy linkage) and is a consequence of strong electron–phonon interactions in the SiO2 crystal structure (Fisher et al. 1990). The CL emission at 2·75–2·80 eV is more or less present in all quartz types (e.g., Mu¨ller 2000, Go¨tze et al. 2001). Thus the defects related to the 2·58 eV, 2·68 eV and 2·79 eV emissions are not only formed during crystal growth, but also induced by ionising radiation and, therefore not indicative for igneous processes. Mu¨ller et al. (2002b) showed that high Ti (>50 ppm) in quartz correlates positively with the intensity of the blue 2·96 eV CL emission. It is still under debate whether Ti is a CL activator or sensitiser (Marshall 1988, Go¨tze 2000). The increasing Al/Ti ratio during magmatic diﬀerentiation indicates that Ti is compatible during the formation of igneous quartz, whereas Al is incompatible. Wark & Watson (2006) proved that Ti4+ –Si4+ substitution is temperature-dependent. The temperature dependence of Ti in quartz explains the higher content in the early-magmatic quartz phenocrysts and the systematic lower concentrations in the younger quartz generations. The relatively stable, blue-dominated CL and the relatively high abundance of Ti in early-magmatic quartz phenocrysts in volcanic rocks and some granites reﬂect crystallisation from a hot and relatively water-poor melt. Pott & McNicol (1971) and Kempe et al. (1999) found that concentration of bound Fe3+ is related to the 1·73 eV CL emission. Some rhyolitic quartz phenocryst margins and the co-genetic ﬁne-grained groundmass quartz show stable red CL at 1·73 eV and elevated Fe concentrations. However, there is no general correlation of total Fe and the 1·73 eV emission intensity found (Mu¨ller 2000). It has been frequently observed that quartz adjacent to Fe-rich minerals (biotite, hornblende, tourmaline, etc.) becomes exponentially enriched in Fe towards the grain contact up to several thousands ppm (Penniston-Dorland 2001; Mu¨ller et al. 2002b, 2003b). However, the diﬀusing ion is preferentially Fe +2 , due to its smaller ion radius. Fe diﬀuses up to 400 m into the quartz, but the Fe-enrichment of quartz does not necessarily result in changing CL properties. Therefore, the total Fe content in quartz cannot be used to trace magmatic processes. Schro¨n et al. (1982) introduced the ternary Ti–Al–Ge diagram to discriminate igneous quartz of diﬀerent origin (‘rhyolitic’, ‘granitic’ and ‘pegmatitic’). They showed that Ge behaves as an incompatible element during magma diﬀerentiation and is a useful quartz ﬁngerprint for magma evolution. Jacamon & Larsen (2009) suggest that the Ge/Ti ratio in quartz is an index of the magmatic evolution of silica-saturated melts. However, the increase of Ge in quartz with increasing diﬀerentiation is only minor until the pegmatitic-pneumatolytic stage, making Ge a useful element for identifying pegmatitic quartz (Schro¨n et al. 1988; Go¨tze et al. 2004). Go¨tze et al. (2005) detected a correlation of the 2·45 eV emission with the concentration of the paramagnetic [GeO4/Li + ]0 centre in pegmatitic quartz.

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES

199

Figure 10 P–T path interpretations for quartz. The grey areas frame thermobarometric results obtained from melt inclusions in felsic igneous rocks from the Erzgebirge, Germany, by Thomas (1994). The data are compiled with experimental PT paths (A, B and C) for haplogranitic magmas with diﬀerent initial water content (Johannes & Holtz 1996). A paths=increasing crystal/melt ratio (decompression crystallisation); B paths=constant crystal melt ratio; C paths=decreasing crystal/melt ratio (decompression melting). To provide an example the change of the quartz crystal proportion in the ascending magma along the path B with the initial conditions of 920(C and 800 MPa is given according to Johannes & Holtz (1996). The quartz portion decreases but feldspar portion increases and the crystals/melt ratio remains constant. Values for the -/-quartz inversion are from Yoder (1950). The dashed solidus lines illustrate the variability of the solidus for diﬀerent granite compositions. See text for explanation of numbers 1 to 7.

6.2. Resorption of quartz phenocrysts Major resorption surfaces observed in early-magmatic phenocrysts in extrusive and plutonic rocks are attributed to profound changes in temperature and magma composition. These surfaces are associated with a concentration step of Ti and commonly have wavy (Fig. 5f), dendritic (Fig. 6a), or granophyric textured overgrowth. Foreign particles adhere at the freshly resorbed surface due to its roughness and the high number of free bonds. The particles cause physical shielding, disturbing the planar growth of the crystal face. The hindered growth may cause growth embayments and wavy textures after major resorption events, which can be ﬁlled later on with melt droplets and foreign crystals (Fig. 5g). Minor quartz resorption reﬂects small changes in the same parameters caused by local magma dynamics (convection). Possible causes of major episodes of quartz resorption were previously discussed by Mu¨ller et al. (2003b) and Mu¨ller et al. (2005) on the example of the Slavkov microgranites and the igneous rocks of the eastern Erzgebirge volcano-plutonic complex. Two possible processes for major resorption of quartz crystals have been extracted from these discussions. First, isothermal decompression due to semi-adiabatic magma ascent and, secondly, mixing and mingling of diﬀerent magmas by injection or recharge. Both processes can be related.

6.2.1. Semi-adiabatic magma ascent. Strong resorption of quartz crystals may occur during semi-adiabatic magma ascent for an energetically closed system (Johannes & Holtz 1996; Fig. 10). Decompression melting during isothermal magma ascent is mainly caused by the decrease of the minimum H2O content of the melt, leading to the increase of the ratio H2Ototal/H2Omelt in the system (Tuttle & Bowen 1958; Carmichael et al. 1974; Sykes & Holloway 1987; Holtz & Johannes 1994; Johannes & Holtz 1996). Experimental data of Johannes & Holtz (1996) obtained for the haplogranite system and illustrated in Figure 10 are used here for better understanding the processes of possible quartz resorption during magma ascent. Figure 10 shows the experimental ascent paths of three haplogranitic magmas with 2 wt.%, 4 wt.%, and 8 wt.% H2O in the melt according Johannes & Holtz (1996). The water-undersaturated magmas start to rise at 800 MPa with 50 wt.% melt and 50 wt.% crystals. During crystallisation along path A, the crystal/melt ratio is increased (decompression crystallisation). Path B represents magma ascent at constant crystal/melt ratio. In the case of an ascending melt along path B with 2 wt.% H2O in the melt and 920(C initial temperature, the relative proportion of quartz and feldspar changes, but the crystal/melt ratio remains constant. Before the solidus is reached, the proportion of quartz is lowered from

200

AXEL MÜLLER ET AL.

Table 3 Increase of quartz crystallisation temperature after major resorption episodes documented by the chemical zoning of quartz phenocrysts. Ti concentrations were determined by EPMA. TiO2–TiO2 activity of magma according to Hayden et al. (2005). T=change of magma temperature; n=number of analyses. Name of igneous rock

Teplice rhyloite (TR2b) Schellerhau granite Altenberg–Frauenstein microgranite Altenberg–Frauenstein microgranite Beucha granite Hammarudda porphyry

Magmatic province

TiO2

Location of resorption surface within phenocrysts

Ti (ppm) of quartz prior to resorption

Ti (ppm) of quartz after resorption

Minimum T after quartz resorption ((C)

Eastern Erzgebirge, Germany Eastern Erzgebirge, Germany Eastern Erzgebirge, Germany Eastern Erzgebirge, Germany NW Saxony, Germany A r land, Finland

0·8 0·8 1 1 1 1

core core core marginal marginal (Fig. 5h) marginal

<22 (n=2) <22 (n=10) <33 (n=10) 59 (n=11) 28 (n=9) 154 (n=16)

58 (n=6) 58 (n=2) 59 (n=11) 120 (n=4) 88 (n=26) 211 (n=7)

+972 +972 +581 +821 +1182 +441

13·5 wt.% to 7·5 wt.% (Fig. 10). On the other hand nearly isothermal decompression (path C), corresponding to adiabatic magma ascent, is accompanied by the release of H2O from the melt in a chemically and energetically closed system. Higher water activities result in melting of the crystals and formation of additional melt (decompression melting). In the C-path example with the initial ascent conditions 800 MPa and 820(C, about 30 vol.% of quartz and feldspar crystals would dissolve between 800 and 50 MPa (Johannes & Holtz 1996). Crystal resorption is much more intense between 300 and 50 MPa than at higher pressures. The simultaneous fractionation of H2O-free phases like quartz and feldspar may result in secondary boiling and magma ascent up to the surface. The magma evolution along path B or C is relatively short when it starts to rise at 710(C, 800 MPa, and 8 wt.% total H2O, as the magma would crystallise completely at 12 km and 8 km, respectively. The magma would become more or less immobile along the A-path and it would crystallise at 600 MPa or even higher pressures. Therefore the development of major resorption textures in quartz requires an episode of about isothermal decompression, as illustrated for paths B and C. The P–T path of rhyolite and granite magmas of the Erzgebirge obtained from microthermometric studies of silicate melt inclusions by Thomas (1992, 1994) is illustrated in Figure 10, together with the generalised quartz textures and populations of quartz (Mu¨ller et al. 2005). The maximum pressure of 780 MPa obtained from melt inclusions represents the minimum crystallisation depth of early magmatic quartz phenocrysts of the Erzgebirge granites and rhyolites (nr. 1 in Fig. 10). It is, however, possible that phenocrysts originate from greater depth. For example, rhombohedric -quartz phenocrysts in rhyolites of the Odenwald (Dossenheim and Weinheim rhyolite, Germany; Fig. 5c) and other occurrences (Bozen rhyolite, Italy) indicate much deeper quartz crystallisation (Flick 1987). Flick (1987) suggested a minimum crystallisation pressure of the -quartz phenocrysts at 1250 to 1400 MPa, assuming a magma temperature between 900(C and 950(C. The P–T-path of the Erzgebirge granites and rhyolites (Thomas 1992, 1994) follows roughly the experimentallydetermined paths A and B for haplogranitic melts with 1·5 wt.% to 2·5 wt.% H2O in melt (Johannes & Holtz 1996). The path of the Erzgebirge magmas is not well documented between 800 and 400 MPa, so that ascent parameters may correspond to the path A, B or even C. A number of phenocryst populations in the Eastern Erzgebirge contain rounded crystal cores, which indicate resorption during the early stage of crystallisation (Fig. 5d; Mu¨ller et al. 2005). This early resorption episode can be attributed to rapid magma ascent or magma mixing. Temporary magma storage probably occurred at a depth of 17–24 km (nr. 2) and in the ductile-

brittle transition zone at a depth of 13–10 km (nr. 3; Mu¨ller et al. 2000, 2005). These stages are documented by signiﬁcant amounts of quartz growth. Decompression crystallisation (path A) must have occurred between 400 and 300 MPa (between nr. 2 and 3) resulting in higher water contents in the melt. Melt inclusion data indicate adiabatic magma ascent (path C) at a depth of 10–4·5 km (nr. 4) that possibly caused quartz phenocryst resorption. An increase of magma temperature could have been caused by mixing with maﬁc magma. Most of the Erzgebirge magmas were emplaced at a shallow depth of 5–2·6 km. Here, the solidifying magmas became supersaturated in volatiles and the pressures only slightly increased along the solidus (nr. 5). Melt inclusion data on the left side of the haplogranite solidus (nr. 6) reﬂect chemistries of highly fractionated magmas enriched in ﬂuorine and other volatiles causing a shift of the solidus. Finally, igneous -quartz crystallised in topaz-bearing albite granites (nr. 7). 6.2.2. Magma mixing. A number of late-Hercynian rhyolites and granites contain phenocrysts with major resorption surfaces, which were aﬀected by magma mixing and mingling (Mu¨ller et al. 2005). Rock-forming minerals in these rocks show textures, which indicate magma mixing, such as plagioclase-mantled K-feldspar, sieve-textured plagioclase and maﬁc micro enclaves (Mu¨ller et al. 2005). The pre-resorption growth zones in these quartz phenocrysts are in some cases smudged and blurred, whereas the boundaries of postresorption growth zones are sharp. The smudging of preresorption growth zones is explained by the redistribution and healing of defect centres in quartz (Mu¨ller et al. 2005). Smudging of growth zoning in quartz is, however, a rare phenomenon amongst the late-Hercynian igneous rocks. The Ti-in-quartz geothermometer of Wark & Watson (2006) provides a tool to determine the quartz crystallisation temperature prior to and after a major phenocryst resorption episode. Table 3 shows calculations of the minimum temperature change after a major resorption episode. The Ti activity TiO2 in the magma was determined according Hayden et al. (2005). Rutile and titanomagnetite are common accessories in the Altenberg–Frauenstein microgranite, Beucha granite, and Hammerrudda porphyry, indicating Ti saturation of the magma. Rutile is rare in the Teplice rhyolite and Schellerhau granite, which implies slight Ti undersaturation; this is conﬁrmed by the calculation method of Hayden et al. (2005). The magma temperature increase in all cases is signiﬁcant. If the resorption is caused by semi-adiabatic magma ascent, the magma moves into a higher and cooler crustal level and the Ti in quartz would not increase after resorption. The ﬁndings support the hypothesis that major resorption of quartz phenocrysts is caused by mixing and mingling with a more maﬁc magma.

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES

201

Thus, a number of strong resorption episodes can be related to magma mixing and mingling, particularly, to those observed in igneous rocks of the eastern Erzgebirge volcano-plutonic complex (Mu¨ller et al. 2005). However, most of the other igneous rocks discussed here show no chemical or textural evidence for magma mixing. These rocks were possibly aﬀected by multiple magma recharge (stepwise magma ascent) resulting in acid-acid magma mixing and mingling. Semi-adiabatic magma ascent theoretically causes major resorption of quartz crystals. At present there is no strong evidence for this theory. However, the injection of a hotter (more maﬁc) magma at the deeper levels of a magma reservoir may lead to higher gas pressures and trigger magma ascent (e.g., Blake et al. 1992, Smith et al. 2004). It is suggested that magma mixing and adiabatic magma ascent go normally hand in hand during the magma evolution, as was demonstrated in the example of the Erzgebirge Granites.

injection or recharge of maﬁc and hotter magmas into the bottom of felsic magma reservoirs may trigger the rise of felsic magma portions. There are still, however, a number of open questions regarding the petrogenetic signiﬁcance of the CL properties and the trace element signature of igneous quartz. In particular, the crystallisation parameters, that control the incorporation of Al into the quartz lattice have to be evaluated by experimental work.

7. Summary

9. References

CL spectra of igneous quartz in late-Hercynian volcanic and plutonic rocks in central and western Europe commonly comprise nine emission bands between 1·4 eV and 3·1 eV. The decay of blue luminescing centres and the generation of red luminescing centres during electron bombardment have been quantiﬁed by applying the kinetic law. Granitic groundmass quartz has more unstable CL, in particular the red 1·85 eV and 1·96 eV emissions, than the CL of early-magmatic quartz phenocrysts. The increase of the 1·85 eV and 1·96 eV emissions during electron radiation is explained by the radiolysis of OHand H2O-defects in quartz. The formation of water-rich disordered micro-domains and the strong increase of red-CL in granitic groundmass quartz reﬂects crystallisation from a water-rich melt in the emplacement level when the melt became rich in water. CL contrasts among growth zones in zoned quartz phenocrysts are essentially related to the distribution of trace elements. Blue luminescing growth zones have high Ti (>50 ppm). High intensity of the relatively stable blue emission at 2·96 eV correlates with high Ti in quartz. As diﬀerentiation proceeds, the Ti concentration in quartz decreases, whereas Al in quartz increases. Thus, Ti behaves compatibly during the evolution of igneous quartz, whereas Al behaves more incompatibly. The more stable, blue-dominated CL and the relative high abundance of Ti in quartz phenocrysts in rhyolites and a number of granites reﬂect higher crystallisation temperatures and a lower crust origin. CL reveals complex growth patterns of early-magmatic quartz phenocrysts in late-Hercynian volcanic and plutonic rocks. Fine-scale (<10 m) oscillatory zoning, large-scale step zoning with strong CL contrasts, skeletal growth, resorption surfaces, and growth embayments are distinguished. A few of late-Hercynian granites contain quartz phenocryst populations similar to phenocrysts in associated rhyolites, indicating that the rhyolite is the extrusive equivalent of the underlying granite. Examples are known from the Eastern Erzgebirge volcano-plutonic complex, e.g. the Schellerhau granites and Teplice rhyolites (Mu¨ller et al. 2005). This phenomenon is presumably much more common, but rhyolites associated with Hercynian granites are either eroded or have not been developed. Phenocrysts in granites are embedded in unzoned anhedral groundmass quartz that makes their identiﬁcation diﬃcult. A number of major resorption episodes can be related to the mixing of magmas with diﬀerent physicochemical properties. There is no strong evidence that the observed major resorption surfaces are caused by semi-adiabatic magma ascent. However, both causes can be related, because the

Aines, R. D. & Rossman, G. R. 1984. Water in minerals? A peak in the infrared. Journal of Geophysical Research 89/B6, 4059–71. Alle`gre, C. J., Provost, A. & Jaupart, C. 1981. Oscillatory zoning: A pathological case of crystal growth. Nature 294, 223–8. Alonso, P. J., Halliburton, L. E., Kohnke, E. E. & Bossoli, R. B. 1983. X-ray induced luminescence in crystalline SiO2. Journal of Applied Physics 54, 5369–75. Armstrong, J. T. 1995. CITZAF: A package of correction programs for the quantitative electron microbeam X-ray analysis of thick polished materials, thin ﬁlms, and particles. Microbeam Analysis 4, 177–200. Bahadur, H. 1994. Sweeping and irradiation eﬀects on hydroxyl defects in crystalline natural quartz. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control 41/6, 820–33. Bambauer, H. U., Brunner, G. O. & Laves, F. 1963. Merkmale des OH-Spektrums alpiner Quarze (3 -Gebiet). Schweizerische Mineralogische und Petrographische Mitteilungen 43, 259–68. Behr, H. J. 1989. Die geologische Aktivita¨t von Krustenﬂuiden. In Behr, H. J., Fyfe, W. S., Hoefs, J., Johannes, W. & Touret, J. L. R. (eds) Gesteinsfluide – Ihre Herkunft und Bedeutung fu¨r geologische Prozesse, Vol. 1, 7–42. Hannover: Niedersa¨chsische Akademie der Geowissenschaften Vero¨ﬀentlichungen. Behr, H.-J. & Frentzel-Beyme, K. 1989. Permeability and paleoporosity in crystalline bedrocks of the Central European basementstudies of cathodoluminescence. In Boden, A. & Eriksson, K. G. (eds) Exploration of the deep continental crust. Deep drilling in crystalline bedrock. Vol. 2, 477–97. Berlin, Heidelberg, New York: Springer. Beus, A. A., Severov, E. A., Sitnin, A. A. & Subbotin, K. D. 1962. Albitized and greisenized granites (apogranites). Moscow: Nauka. [In Russian.] Blake, S., Wilson, C. J. N., Smith, I. E. M. & Walker, G. P. L. 1992. Petrology and dynamics of the Waimihia mixed magma eruption, Taupo Volcano, New Zealand. Journal of the Geological Society, London 149, 193–207. Blankenburg, H.-J., Go¨tze, J. & Schulz, J. 1994. Quarzrohstoﬀe. Leipzig-Stuttgart, Germany: Deutscher Verlag fu¨r Grundstoﬃndustrie. Botis, S., Nokhrin, S. M., Pan, Y., Xu, Y. & Bonli, T. 2005. Natural radiation-induced damage in quartz. I. Correlations between cathodoluminescence colors and paramagnetic defects. Canadian Mineralogist 43, 1565–80. Breiter, K., Mu¨ller, A., Leichmann, J. & Gabasˇova, A. 2005. Textural and chemical evolution of a fractionated granitic system: the Podlesı´ stock, Czech Republic. Lithos 80, 323–45. Carmichael, I. S. E., Turner, F. J. & Verhoogen, J. 1974. Igneous petrology. New York: McGraw-Hill. Charlier, B. L. A., Wilson, C. J. N., Lowenstern, J. B., Blake, S., van Calsteren, P. W. & Davidson, J. P. 2005. Magma generation at a large, hyperactive silicic volcano (Taupo, New Zealand) revealed by U–Th and U–Pb systematic in zircons. Journal of Petrology 46, 3–32. Davidson, J., Tepley III, F., Palacz, Z. & Meﬀan-Main, S. 2001. Magma recharge, contamination and residence times revealed by in situ laser ablation analysis of feldspar in volcanic rocks. Earth and Planetary Science Letters 184, 427–42.

8. Acknowledgements This study was supported by the Deutsche Forschungsgemeinschaft (MU 1717/2–1), the Natural History Museum, London and the Geological Survey of Norway. We are grateful to Marian Holness, Jon Davidson and John D. Clemens for constructive reviews and discussions.

202

AXEL MÜLLER ET AL.

Davidson, J., Morgan D. J., Charlier, B. L. A., Harlou, R. & Hora, J. M. 2007. Microsampling and isotopic analysis of igneous rocks: Implications for the study of magmatic systems. Annual Reviews of Earth and Planetary Sciences 35, 273–311. D’Lemos, R. S., Kearsley, A. T., Pembroke, J. W., Watt, G. R. & Wright, P. 1997. Complex quartz growth histories in granite revealed by scanning cathodoluminescence techniques. Geological Magazine 134, 459–552. Fenn, P. M. 1977. The nucleation and growth of alkali feldspars from hydrous melts. Canadian Mineralogist 15, 135–61. Fenn, P. M. 1986. On the origin of graphic granite. American Mineralogist 71, 325–30. Fisher A. J., Hayes W. & Stoneham A. M. 1990. Structure of the self-trapped exciton in quartz. Physical Reviews Letters 64, 2667–70. Flick, H. 1987. Geotektonische Verknu¨pfung von Plutonismus und Vulkanismus im su¨dwestdeutschen Variscicum. Geologische Rundschau 76, 699–707. Fowler, A. D. 1990. Self-organized mineral textures of igneous rocks: the fractal approach. Earth Sciences Reviews 29, 47–55. Gorton, N. T., Walker, G. & Burley, S. D. 1996. Experimental analysis of the composite blue CL emission in quartz. Journal of Luminescence 72–74, 669–71. Go¨tze, J. 2000. Cathodoluminescence microscopy and spectroscopy in applied mineralogy. Freiberger Forschungshefte C 485, 1–128. Go¨tze, J., Plo¨tze, M., Fuchs, H. & Habermann, D. 1999. Defect structure and luminescence behaviour of agate – results of electron paramagnetic resonance (EPR) and cathodoluminescence (CL) studies. Mineralogical Magazine 63/2, 149–63. Go¨tze, J., Plo¨tze, M. & Habermann, D. 2001. Origin, spectral characteristics and practical applications of the cathodoluminescence (CL) of quartz – a review. Mineralogy and Petrology 71, 225–50. Go¨tze, J., Plo¨tze, M. Graupner, T., Hallbauer, D. K. & Bray, C. J. 2004. Trace element incorporation into quartz: A combined study by ICP-MS, electron spin resonance, cathodoluminescence, capillary ion analysis, and gas chromatography. Geochimica et Cosmochimica Acta 68, 3741–59. Go¨tze, J., Plo¨tze, M. & Trautmann, T. 2005. Structure and luminescence characteristics of quartz from pegmatites. American Mineralogist 90, 13–21. Gritsenko, B. P. & Lisitsyn, V. M. 1985. Intrinsic short-lived defects in quartz. Soviet Physics – Solid State 27, 1330–1. Gurbanov, A. G., Chernukha, D. G., Koshchug, D. G., Kurasova, S. P. & Fedyushchenko, S. V. 1999. EPR spectroscopy and geochemistry of rock-forming quartz as an indicator of the superimposed processes in rocks of igneous associations of various ages in the Greater Caucasus. Geochemistry International 37, 519–604. Hayden, L. A., Watson, E. B. & Wark, D. A. 2005. Rutile saturation and TiO2 diﬀusion in hydrous siliceous melts. EOS Transactions, American Geophysical Union 86, Fall meeting supplement, MR13A-0076. Heggie, M. I. 1992. A molecular water pump in quartz dislocations. Nature 355, 337–9. Holtz, F. & Johannes, W. 1994. Maximum and minimum water contents of granitic melts: implications for chemical and physical properties of ascending magmas. Lithos 32, 149–59. Itoh, C., Suzuki, T. & Itoh, N. 1990. Luminescence and defect formation in undensiﬁed and densiﬁed amorphous SiO2. Physical Reviews B 41/6, 3794–9. Jacamon, F. & Larsen, R. B. 2009. Trace element evolution of quartz in the charnockitic Kleivan granite, SW Norway: the Ge/Ti ratio of quartz as an index of igneous diﬀerentiation. Lithos 107, 281–91. Jani, M., Halliburton, L. E. & Kohnke, E. E. 1983. Point defects in crystalline SiO2: thermally stimulated luminescence above room temperature. Journal of Applied Physics 54, 6321–8. Johannes, W. & Holtz, F. 1996. Petrogenesis and experimental petrology of granitic rocks. Berlin, Heidelberg, New York: Springer. Kats, A. 1962. Hydrogen in alpha-quartz. Philips Research Reports 17, 133–279. Kempe, U., Go¨tze, J., Dandar, S. & Habermann, D. 1999. Magmatic and metasomatic processes during formation of the Nb–Zr–REE deposits from Khaldzan Buregte (Mongolian Altai): Indications from a combined CL-SEM study. Mineralogical Magazine 63, 165–77. Kirkpatrick, R. J. 1981. Kinetics of crystallization of igneous rocks. Reviews in Mineralogy 8, 321–98. Washington DC: Mineralogical Society of America. Kovalenko, V. I. 1977. Petrology and geochemistry of rare-metal granitoids. Novosibirsk: Nauka. [In Russian.]

Kronenberg, A. K., Kirby, S. H., Aines, R. D. & Rossmann, G. R. 1986. Solubility and diﬀusional uptake of hydrogen in quartz at high water pressures: implications for hydrolytic weakening in the laboratory and within the earth. Tectonophysics 172, 255–71. Laemmlein, G. 1930. Korrosion und Regeneration der PorphyrQuarze. Zeitschrift fu¨r Kristallographie 75, 109–27. Larsen, R. B., Polve´, M. & Juve, G. 2000. Granite pegmatite from Evje-Iveland: trace element chemistry and implications for the formation of high-purity quartz. Norges geologiske undersøkelse Bulletin 436, 57–65. Larsen, R. B., Henderson, I., Ihlen, P. M. & Jacamon, F. 2004. Distribution and petrogenetic behaviour of trace elements in granitic quartz from South Norway. Contributions to Mineralogy and Petrology 147, 615–28. Luﬀ, B. J. & Townsend, P. D. 1990. Cathodoluminescence of synthetic quartz. Journal of Physics: Condensed Matter 2, 8089–97. Lyakhovich, V. V. 1972. Trace elements in rock-forming minerals of granitoids. Moscow: Izd. Nedra. 200 pp. [In Russian.] Mashima, H. 2004. Time scale of magma mixing between basalt and dacite estimated for the Saga–Futagoyama volcanic rocks in northwest Kyushu, southwest Japan. Journal of Volcanology and Geothermal Research 131, 333–49. Marshall, D. J. 1988. Cathodoluminescence of geological materials. Winchester, Massachusetts: Allen & Unwin Inc. Mosenfelder, J. L. 2000. Pressure dependence of hydroxyl solubility in coesite. Physics and Chemistry of Minerals 27, 610–17. Mu¨ller, A. 2000. Cathodoluminescence and characterisation of defect structures in quartz with applications to the study of granitic rocks. Ph.D. Thesis, University of Go¨ttingen, Germany. Mu¨ller, A., Seltmann, R. & Behr, H. J. 2000. Application of cathodoluminescence to magmatic quartz in a tin granite – case study from the Schellerhau Granite Complex, Eastern Erzgebirge, Germany. Mineralium Deposita 35, 169–89. Mu¨ller, A., Kronz, A. & Breiter, K. 2002a. Trace elements and growth pattern in quartz: a ﬁngerprint of the evolution of the subvolcanic Podlesı´ Granite System (Krusˇne´ Hory, Czech Republic). Bulletin of Czech Geological Survey 77, 135–45. Mu¨ller, A., Lennox, P. & Trzebski, R. 2002b. Cathodoluminescence and micro-structural evidence for crystallisation and deformation processes of granites in the Eastern Lachlan Fold Belt (SE Australia). Contributions to Mineralogy and Petrology 143, 510–24. Mu¨ller, A., Wiedenbeck, M., Van den Kerkhof, A. M., Kronz, A. & Simon, K. 2003a. Trace elements in quartz – a combined electron microprobe, secondary ion mass spectrometry, laser-ablation ICP-MS, and cathodoluminescence study. European Journal of Mineralogy 15, 747–63. Mu¨ller, A., Rene´, M., Behr, H.-J. & Kronz, A. 2003b. Trace elements and cathodoluminescence of igneous quartz in topaz granites from the Hub Stock (Slavkovsky´ Les Mts, Czech Republic). Mineralogy and Petrology 79, 167–91. Mu¨ller, A., Breiter, K., Seltmann, R. & Pe´cskay, Z. 2005. Quartz and feldspar zoning in the Eastern Erzgebirge pluton (Germany, Czech Republic): evidence of multiple magma mixing. Lithos 80, 201–27. Mu¨ller, A., Seltmann, R., Halls, C., Jeﬀries, T., Dulski, P., Spratt, J. & Kronz, A. 2006a. The Land’s End granite, Cornwall: the evolution of a composite and mineralised pluton. Ore Geology Reviews 28, 329–67. Mu¨ller, A., Thomas, R., Wiedenbeck, M., Seltmann, R. & Breiter, K. 2006b. Water content of granitic melts from Cornwall and Erzgebirge: A Raman spectroscopy study of melt inclusions. European Journal of Mineralogy 18, 429–40. Mu¨ller, A., Ihlen, P. M. & Kronz A. 2008. Quartz chemistry in polygeneration Sveconorwegian pegmatites, Froland, Norway. European Journal of Mineralogy 20, 447–63. Mu¨ller, A. & Seltmann, R. 1999. The genetic signiﬁcance of snowball quartz in high fractionated tin granites of the Krusˇne Hory/ Erzgebirge. In Stanley, C. J., Rankin, A. H., Bodnar, R. J., Naden, J., Yardley, B. W. D., Criddle, A. J., Hagni, R. D., Gize, A. P., Pasava, J., Fleet, A. J., Seltmann, R., Halls, C., Stemprok, M., Williamson, B., Herrington, R. J., Hill, R. E. T., Prichard, H. M., Wall, F., Williams, C. T., McDonald, I., Wilkinson, J. J., Cooke, D., Cook, N. J., Marshall, B. J., Spry, P., Khin Zaw, Meinert, L., Sundblad, K., Scott, P., Clark, S. H. B., Valsami-Jones, N. J., Beukes, E., Stein, H. J., Hannah, J. L., Neubauer, F., Blundell, D. J., Alderton, D. H. M., Smith, M. P., Mulshaw, S. & Ixer, R. A. (eds) Mineral deposits: processes to processing 1, 409–12. Rotterdam: Balkema.

QUARTZ IN LATE-HERCYNIAN GRANITES AND RHYOLITES Neuser, R. D., Bruhn, F., Go¨tze, J., Habermann, D. & Richter, D. K. 1995. Kathodolumineszenz: Methodik und Anwendung. Zentralblatt fu¨r Geologie und Pala¨ontologie Teil 1 1/2, 287–306. Owen, M. R. 1988. Radiation-damage halos in quartz. Geology 16, 529–32. Penniston-Dorland, S. C. 2001. Illumination of vein quartz textures in a porphyry copper ore deposits using scanned cathodoluminescence: Grasberg Igneous Complex, Irian Jaya, Indonesia. American Mineralogist 86, 652–66. Pearce, T. H. & Kolisnik, A. M. 1990. Observations of plagioclase zoning using interference imaging. Earth Science Reviews 29, 9–26. Peppard, B. T., Steele, I. M., Davis, A. M., Wallace, P. J. & Anderson, A. T. 2001. Zoned quartz phenocrysts from the rhyolitic Bishop Tuﬀ. American Mineralogist 86, 1034–52. Perny, B., Eberhardt, P., Ramseyer, K., Mullis, J. & Pankrath, R. 1992. Microdistribution of aluminium, lithium and sodium in a quartz: possible causes and correlation with shored lived cathodoluminescence. American Mineralogist 77, 534–44. Pichavant, M. 1981. An experimental study of the eﬀect of boron on water-saturated haplogranite at 1 kbar pressure: geological applications. Contributions to Mineralogy and Petrology 76, 430–9. Picouet, P. A. 1999. Cathodoluminecence spectroscopy and the orientation of a hydrothermal quartz crystal. Schweizerische Mineralogische und Petrographische Mitteilungen 79, 455–60. Pott, G. T. & McNicol, B. D. 1971. Spectroscopy study of the coordination and valence of Fe and Mn ions in and on the surface of aluminas and silicas. Discussions of the Faraday Society 52, 121–31. Poutiainen, M. & Scherbakova, T. F. 1998. Fluid and melt inclusion evidence for the origin of idiomorphic quartz crystals in topazbearing granite from the Salmi batholith, Karelia, Russia. Lithos 44, 141–51. Rajpoot, G. S. 1991. Controversies in geotectonics of Hercynian Europe. Prague: Czech Geological Survey. Ramseyer, K., Baumann, J., Matter, A. & Mullis, J. 1988. Cathodoluminescence colours of -quartz. Mineralogical Magazine 52, 669–77. Remond, G., Cesbron, F., Chapoulie, R., Ohnenstetter, D., RouquesCarmes, C. & Schvoerer, M. 1992. Cathodoluminescence applied to the microcharacterization of mineral materials: a present status in experimentation and interpretation. Scanning Microscopy 6/1, 23–69. Rink, W. J., Rendell, H., Marseglia, E. A., Luﬀ, B. J. & Townsend, P. D. 1993. Thermoluminescence Spectra of igneous quartz and hydrothermal quartz. Physics and Chemistry of Minerals 20, 353–61. Rovetta, M. R., Blacic, J. D., Hervig, R. L. & Holloway, J. R. 1989. An experimental study of hydroxyl in quartz using infrared spectroscopy and ion microprobe techniques. Journal of Geophysical Research 94, 5840–50. Ruﬃni, R., Borghi, A., Cossio, R., Olmi, F. & Vaggelli, G. 2002. Volcanic quartz growth zoning identiﬁed by cathodoluminescence and EPMA studies. Mikrochemica Acta 139, 151–8. Rusk, B. & Reed, M. 2002. Scanning electron microscope– cathodoluminescence analysis of quartz reveals complex growth histories in veins from the Butte porphyry copper deposit, Montana. Geology 30, 727–30. Schneider, J. 1993. Das lumineszenzaktive Strukturinventar von Quarzpha¨nokristen in rhyolithen. Go¨ttinger Arbeiten zur Geologie und Pala¨ontologie 60, 1–81. Schro¨n, W., Schma¨dicke, E., Thomas, R. & Schmidt, W. 1988. Geochemische Untersuchungen an Pegmatitquarzen. Zeitschrift fu¨r Geologische Wissenschaften 16, 229–44. Sigel, G. H. Jr. & Marrone, M. J. 1981. Photoluminescence in as-drawn and irradiated silica optical ﬁbres: An assessment of the role of nonbridging oxygen defect centres. Journal of Non Crystalline Solids 45, 235–47. Smith, V. C., Shane, P. & Nairn, I. A. 2004. Reactivation of a rhyolitic magma body by new rhyolitic intrusion before the 15·8 ka Rotora eruptive episode: implications for magma storage in the Okataina Volcanic Centre, New Zealand. Journal of the Geological Society, London 161, 757–72. Sonyushkin, V. E., Sukhorukov, Y. T. & Scherbakova, T. F. 1991. P–T environment of crystallization of quartz in granites of the Salmi batholith, Russian Karelia. In Haapala, I. & Ra¨mo¨, O. T. (eds) Symposium on Rapakivi Granites and Related Rocks. Abstracts, Geological Survey of Finland Guide 34, 47. Sprunt, E. S. 1981. Causes of quartz cathodoluminescence colors. Scanning Electron Microscopy 1981, 525–35.

203

Sprunt, E. S. & Nur, A. 1979. Microcracking and healing in granites: New evidence from cathodoluminescence. Science 205, 495–7. Stenina, N. G., Bazarov, L. S., Shcherbakova, M. Y. & Mashkovtsev, R. I. 1984. Structural state and diﬀusion of impurities in natural quartz of diﬀerent genesis. Physics and Chemistry of Minerals 10, 180–6. Stevens Kalceﬀ, M. A., Phillips, M. R., Moon, A. R. & Kalceﬀ, W. 2000. Cathodoluminescence microcharacterization of silicon dioxide polymorphs. In Pagel, M., Barbin V., Blanc P. & Ohnenstetter, D. (eds) Cathodoluminescence in Geosciences, 193–224 Berlin, Heidelberg, New York: Springer. Stevens Kalceﬀ, M. A. & Phillips, M. R. 1995. Cathodoluminescence microcharacterization of the defect structure of quartz. Physical Reviews B 52, 3122–34. Suttner, L. & Leininger, R. K. 1972. Comparison of the trace element content of plutonic, volcanic and metamorphic quartz from Southwestern Montana. Geological Society of America Bulletin 83, 1855–62. Swanson, S. E. 1977. Relation of nucleation and crystal-growth to the development of granitic textures. American Mineralogist 62, 966–78. Sykes, M. L. & Holloway, J. R. 1987. Evolution of granitic magmas during ascent: a phase equilibrium model. In Mysen, B. O. (ed.) Magmatic processes: Physicochemical Principles. Geochemical Society, Special Publication 1, 447–61. St Louis, Missouri: Geochemical Society. Thomas, R. 1992. Results of investigations on melt inclusions in various magmatic rocks from the northern border of the Bohemian Massif. In Kukal, Z. (ed.) Proceedings of the 1st International Conference about the Bohemian Massif. Prague: Czech Geological Survey. Thomas, R. 1994. Fluid evolution in relation to the emplacement of the Hercynian granites in the Erzgebirge region: a review of the melt and ﬂuid inclusion evidence. In Seltmann, R., Mo¨ller, P. & Ka¨mpf, H. (eds) Metallogeny of Collisional Orogens. Prague: Czech Geological Survey. Tuttle, O. F. & Bowen, N. L. 1958. Origin of granite in the light of experimental studies in the system NaAlSi3O8–KalSi3O8–SiO2– H2O. Geological Society of America Memoir 74, 1–153. Valley, J. W. & Graham, C. M. 1996. Ion microprobe analysis of oxygen isotope ratios in quartz from Skye granite: healed microcracks, ﬂuid ﬂow, and hydrothermal exchange. Contributions to Mineralogy and Petrology 124, 225–34. Van den Kerkhof, A. M., Kronz, A. & Simon, K. 2001. Trace element redistribution in metamorphic quartz and ﬂuid inclusion modiﬁcation: observations by cathodoluminescence. XVI ECROFI, Porto 2001, Departomento Geologica Memo´ria 7, 447–50. Van den Kerkhof, A. M., Kronz, A., Simon, K. & Scherer, T. 2004. Fluid-controlled quartz recovery in granulite as revealed by cathodoluminescence and trace element analysis (Bamble sector, Norway). Contributions to Mineralogy and Petrology 146, 637–52. Van den Kerkhof, A. M. & Grantham, G. H. 1999. Metamorphe charnockite in contact aureoles around intrusive enderbite from Natal, South Afrika. Contributions to Mineralogy and Petrology 137, 115–32. Van den Kerkhof, A. M. & Hein, U. 2001. Fluid inclusion petrography. Lithos 55, 27–47. Van den Kerkhof, A. M. & Mu¨ller, A. 1999. Fluid inclusion re-equilibration and trace element redistribution in quartz: observations by cathodoluminescence microscopy. ECROFI XV 1999 Abstracts and Program, Potsdam, Terra Nostra 99/6, 161–2. Vollbrecht, A., Rust, S. & Weber, K. 1991. Development of microcracks in granites during cooling and uplift: examples from the Hercynian basement in NE Bavaria, Germany. Journal of Structural Geology 13, 787–99. Vollbrecht, A., Olesen, N. O., Schmidt, N. H. & Weber, K. 1994. Crystallographic microcrack orientation in quartz from a granite – a combined ECP/U stage study. In Bunge, H. J., Siegesmund, S., Skrotzki, W. & Weber, K. (eds) Textures of geological materials, 345–52. Oberursel: DGM Informationsgesellschaft Verlag. Wark, D. A., Hildreth, W., Spear, F. S., Cherniak, D. J. & Watson, E. B. 2007. Pre-eruption recharge of the Bishop magma system. Geology 35, 235–8. Wark, D. A. & Watson, E. B. 2006. TitaniQ: a titanium-in-quartz geothermometer. Contributions to Mineralogy and Petrology 152, 743–54. Watt, G. R., Wright, P., Galloway, S. & McLean, C. 1997. Cathodoluminescence and trace element zoning in quartz phenocrysts and xenocrysts. Geochimica et Cosmochimica Acta 61, 4337–48.

204

AXEL MÜLLER ET AL.

Wiebe, R. A., Wark, D. A. & Hawkins, D. P. 2007. Insights from quartz cathodoluminescence zoning into crystallization of the Vinalhaven granite, coastal Maine. Contributions to Mineralogy and Petrology 154, 439–53. Yang, X. H., Townsend, P. D. & Holgate, S. A. 1994. Cathodoluminescence and depth proﬁles of tin in ﬂoat glass. Journal of Physics D 27, 1757–62.

Yoder, H. R. 1950. High-low quartz inversion up to 10,000 bars. Transactions, American Geophysical Union 31, 827–35.

MS received 3 December 2007. Accepted for publication 2 June 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 205–218, 2010 (for 2009)

Trace element constraints on mid-crustal partial melting processes – A garnet ionprobe study from polyphase migmatites (Damara orogen, Namibia) C. Jung1, S. Jung2,3, E. Hellebrand2,4 and E. Hoﬀer1 1

Fachbereich Geowissenschaften, Philipps Universita¨t Marburg, 35032 Marburg, FRG

2

Max-Planck-Institut fu¨r Chemie, Abt. Geochemie, Postfach 3060, 55020 Mainz, FRG

3

Department Geowissenschaften, Mineralogisch-Petrographisches Institut, Universita¨t Hamburg, 20146 Hamburg, FRG Email: [email protected]

4

SOEST – Department of Geology and Geophysics, University of Hawaii, USA

ABSTRACT: Trace element abundances in garnet from a polyphase migmatite were measured by secondary ion mass spectrometry (SIMS) in order to identify some of the eﬀective variables on the trace element distribution between garnet and melanosome or leucosome. In general, garnet is zoned with respect to REE, in which garnet cores are enriched by a factor of 2–3 relative to the rims. For an inclusion-rich garnet from the melanosome, equilibrium distribution following a simple Rayleigh fractionation is responsible for the decreasing concentrations in REE from core to rim. Inclusionpoor garnet from the same melanosome located in the vicinity of the leucosomes shows distinct enrichment and depletion patterns for REE from core to rim. These features suggest disequilibrium between garnet and the host rock which, in this case, could have been an in-situ derived melt. This would probably indicate a period of open-system behaviour at a time when the garnet, originally nucleated in the metamorphic environment reacted with the melt. In addition, non-gradual variation in trace element abundances between core and rim may suggest variable garnet growth rates. Inclusion-free garnet from the leucosome, interpreted to have crystallised in the presence of a melt, has a small core with high REE abundances and a broad rim with lower REE abundances. Here, crystal-liquid diﬀusion-controlled partitioning is a likely process to explain the trace element variation. KEY WORDS: in situ partial melting, granites, migmatites, trace element composition Garnet is the major mineral used for the reconstruction of P–T paths because it is involved in a number of reactions useful for thermobarometry (e.g. Essene 1989). Consequently, the major element zoning of garnet has been widely used to investigate the P–T conditions of regional metamorphic terranes (Tracy 1982; Spear & Selverstone 1983; Spear et al. 1984). The chemical signatures of the early stages of regional metamorphism are often preserved up to the lower amphibolite facies, due to the slow diﬀusivities of the major cations Fe, Mg, Mn and Ca. However, prograde chemical zoning is rare in highgrade garnet of amphibolite- to granulite-facies terranes, due to enhanced intracrystalline diﬀusion at higher temperatures (Spear & Florence 1992). On the other hand, it has been demonstrated that trace element zoning in garnet is often more pronounced than major element zoning (Hickmott et al. 1987; Hickmott & Shimizu 1990; Hickmott & Spear 1992; Spear & Kohn 1996; Chernoﬀ & Carlson 1999; Otamendi et al. 2002). This implies that trace elements may potentially be less sensitive to chemical changes in rocks than major elements. Accurate interpretation of major and trace element zoning in garnet can therefore lead to signiﬁcantly improved understanding of the evolutionary path of metamorphism and hence of the tectonic evolution of mountain belts.

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016107

Prograde heating of crustal segments will ultimately result in ﬂuid-absent partial melting of metasedimentary rocks, producing a granitic melt and an aluminous residue in which the restitic mineral assemblage will control the trace element concentration of the melt. The mineralogical composition of the residual assemblage is dependent on the nature of the melting reaction and the pressure at which melting occurs. Garnet is a common refractory mineral phase that may be present in the high-grade metasedimentary protolith, or a product of incongruent melting reactions at higher metamorphic grade. A detailed knowledge of the distribution of trace elements between garnet, metamorphic host rock or melt is therefore critical for trace element modelling of anatectic crustal melts. In order to identify some of the eﬀective variables on the trace element distribution between garnet and the host rock, this study presents trace element data on garnet measured by SIMS from a polyphase migmatite terrane from the Damara orogen (Namibia). It is shown that trace element variation during garnet growth during regional metamorphism and melting provides additional constraints on the garnet-forming process with important implications for the geochemical signature of granitic melts.

206

C. JUNG ET AL.

Figure 1 Generalised geological map showing the study area within the transition zone between the lower part of the southern Central Zone and Southern Zone of the Damara orogen, Namibia. Abbreviations in inset: KZ=Kaoko Zone; NP=Northern Platform; NZ=Northern Zone; nCZ=northern central Zone; sCZ=southern Central Zone; SZ=Southern Zone; SMZ=Southern Margin Zone. Isograd map (Hartmann et al. 1983) gives the distribution of pan-African regional metamorphic isograds within the southern and central Damara orogen. Isograds: (1) biotite-in; (2) garnet-in; (3) staurolite-in; (4) kyanite-in; (5) cordierite-in; (6) andalusite <--->sillimanite; (7) sillimanite-in according to staurolite-breakdown; (8) partial melting due to: muscovite+plagioclase+quartz+H2O <--->melt+sillimanite; (9) K-feldspar+cordierite-in; (10) partial melting due to: biotite +K-feldspar+plagioclase+quartz+cordierite <--->melt+garnet.

1. Geological setting The Damara orogen of Namibia comprises a deeply eroded section of a Pan-African mobile belt that can be divided into a N–S-trending coastal branch, the Kaoko belt, and a NE–SWtrending intracontinental branch (see inset to Fig. 1). This mobile belt has been divided into several zones based mainly on stratigraphy, metamorphic grade, structure and geochronology (Miller 1983). The intrusive rocks record ages between 750 Ma and 480 Ma and crop out over an area of approximately 75 000 km2 (Fig.1). Granites are the most common igneous rock type (c. 94%) and the remaining intrusions are equally divided between diorite and tonalite/granodiorite (Miller 1983). Pre-Damara basement gneisses are overlain by various metasedimentary sequences containing quartzose sandstones, arkoses, schists and calc-silicate rocks, marble, conglomerate, metamorphosed glaciogenic diamictites, banded iron-stones, Al-rich metapelite and carbonates. In the Central Zone (Fig. 1), the metamorphic grade increases from east to west, reaching high-grade conditions with local partial melting in the coastal area (Hartmann et al. 1983). Early maximum estimates for the peak metamorphic temperatures were ca. 645(C at 31 kbar obtained on impure marbles (Puhan 1983) and ca. 650(C based on oxygen isotope fractionation in a variety of metasedimentary and metaigneous rocks (Hoernes & Hoﬀer 1979). More recently, pressure–temperature estimates were substantially revised indicating low-pressure high-temperature granulite facies conditions with temperatures between 700–750(C at 5–6 kbar

(Masberg et al. 1992; Jung & Mezger 2003; Ward et al. 2008). These new p–t estimates are generally considered to be too low to cause widespread biotite dehydration melting, however, migmatisation is a local phenomenon and is not accompanied by large volumes of newly-formed melt. There is some disagreement as to whether the initial steps of melting were caused by muscovite-dehydration melting, since metamorphic K-feldspar is a common mineral in high-grade metapelites (Jung et al. 1999; Jung & Mezger 2003). On the other hand, initial partial melting under water-saturated conditions is a possibility; in this case the water may have come from some of the plutons associated with the migmatites. One major step forward in the interpretation of the metamorphic history of the Damara orogen is provided by Ward et al. (2008), who showed that ﬂuid-assisted partial melting involving biotite breakdown is a possibility to explain some of the migmatites in the central Damara orogen. The large amount of sizeable plutons observed in the Central Damara orogen originated by partial melting processes in the deeper crust where temperatures in excess of 800–850(C are not unlikely. To the south-east, there is a gradation into the Okahandja Lineament Zone that separates the Central Zone from the Southern Zone (Fig. 1). In the Southern Zone, regional metamorphism is characterised by a Barrovian-type sequence, with a general increase in the metamorphic grade from south to north. The metamorphic conditions range from low to medium pressures and reached up to 8 kbar at maximum temperatures of 600(C. There is evidence that suggests that

TRACE ELEMENT CONSTRAINTS – A GARNET IONPROBE STUDY

207

Figure 2 Sketch of the migmatite sample from the Davetsaub area showing the diﬀerent lithological units.

regional metamorphism in the Southern Zone is more or less coeval with regional metamorphism in the Central Zone. This interpretation is based on U–Pb and Pb–Pb garnet and staurolite ages of 55010 Ma and 48020 Ma (Jung 2005), similar to the range of U–Pb monazite and Sm–Nd garnet ages obtained from metapelites and migmatites from the Central Zone. U–Pb garnet and staurolite ages must be interpreted with caution, since any incorporated high-U/Pb microinclusion may totally cover the initial radiogenic isotope composition of garnet and staurolite. However, in the present case, 206Pb/ 204 Pb ratios of garnet and staurolite are <50, suggesting that high U/Pb inclusions do not play a major role here and the ages, although associated with a relatively large error, date the time of garnet and staurolite growth. Rb–Sr and K–Ar biotite ages indicate cooling to 350–300(C between 480 Ma and 460 Ma (Miller 1983). More recently, Gray et al. (2006) obtained Ar–Ar isotope data on white mica and biotite, indicating cooling to 300–350(C between 495 Ma and 480 Ma. Intrusion of large volumes of granitic rocks are absent even in the highest grade zone of the Southern Damara orogen. The formation of the migmatites of Davetsaub and Nomatsaus (22·4(S/16·3(E, Fig. 1) was originally thought to be related to the intrusion of the peak-metamorphic postdeformational Donkerhoek granodiorite (Nieberding 1976; Saywer 1981). Since the granodiorite has no extensive thermal aureole, the country rocks are assumed to have been still hot at the time of emplacement. Based on the geological map (Miller 1983), isotopic investigations by Kukla (1993), and detailed ﬁeld investigations by the present authors, it is more likely that the Davetsaub area represents a portion of the Damaran metasedimentary rocks where intrusion of older granites produced various types of migmatites. Kukla (1993) presented structural evidence that the anatectic event that generated the migmatites occurred relatively late within the deformational

history of the Damara orogen, but clearly predates the intrusion of the Donkerhoek granodiorite.

2. Description of the migmatite Macroscopic and mesoscopic descriptions of the migmatites from the Davetsaub and Nomatsaus areas have been published by Kukla (1993), Kukla et al. (1991) and Jung et al. (1998). Additional information about the mineralogical composition of the diﬀerent sub-domains of the migmatite (mesosome, melanosome, leucosome) is given in Jung et al. (1998). Metasedimentary and migmatitic rocks outcropping at the Davetsaub and Nomatsaus areas (Fig. 1) are ﬁne to mediumgrained dark-grey to dark-brown pelitic gneisses with a fabric deﬁned by alternating biotite-bearing and biotite-poor domains. Most pelitic gneisses are impregnated by pegmatitic material, as well as concordant to discordant leucosomes. Banded migmatites are commonly folded, and most of the pegmatites cut the main foliation. Mesosomes (nomenclature according to Brown 1983) and leucosomes are well segregated and locally separated by biotite–garnet–cordierite-bearing melanosomes. Some leucocratic layers contain isolated xenoliths with the mineral assemblage quartz+plagioclase+ sillimanite garnetK-feldsparbiotite (Fig. 2). In some areas, the proportion of leucocratic material dominates, so that some melanosomes appear as rafts surrounded by leucosome material and the resulting rock is a diatexite. In the biotite-bearing domains of the mesosomes and melanosomes, the fabric is deﬁned by aligned biotite ﬂakes, biotite–sillimanite intergrowths, xenomorphic garnet porphyroblasts with inclusions of biotite, sillimanite, plagioclase and spinel and poikilitic, elongated, partly pinitised cordierite porphyroblasts with inclusions of sillimanite. Garnet has abundant (<30 m) inclusions of zircon and monazite. In the biotite-poor domains of

208

C. JUNG ET AL.

some mesosomes, ribboned quartz aggregates and relict plagioclase porphyroblasts dominate the fabric. Textural relations preserved within the pelitic migmatites suggest that the assemblage quartz+plagioclase+biotite+ cordierite(I) +sillimanite was developed at an early stage in the metamorphic history. Subsequently, the growth of garnet spinel at the expense of biotite and/or cordierite can be observed. Potential coexistence of spinel and quartz could be interpreted to be indicative of T>800(C during an earlier portion of the metamorphic path, although the signiﬁcant gahnite component in the spinel suggests stability with quartz could have been possible at slightly lower temperatures (Vielzeuf 1983). There is no petrographic evidence for the coexistence of spinel and quartz, judging from the occurrence of spinel exclusively in garnet and, therefore, high grade metamorphism and partial melting must have occurred prior to the formation of the observed mineral assemblage, when aH2O was higher. Melting could have been initiated at ca. 670(C at 3–5 kbar, producing small amounts of liquid while eﬃciently purging the remaining rock of H2O (Thompson 1982). This initial melting step could have been watersaturated, in which the water needed for H2O-saturated melting was derived from the nearby plutons. It is, however, diﬃcult or even impossible to prove whether initial melting was water-saturated in rocks that have evolved to higher temperatures, and especially in those rocks that have lost melt. Unfortunately, no precise radiometric age determinations from all plutons in this area are available. In some layers, relict garnet porphyroblasts are often rimmed and partly replaced by coronas of cordierite(II) suggesting decompression later in the metamorphic history. Garnet in the leucocratic layers of the migmatites is idioblastic and inclusion-free and remained apparently unreacted, or is only locally replaced by an aggregate consisting of green biotite and Ab-rich plagioclase. Fresh cordierite, sometimes twinned, occurs as large sub-idiomorphic grains, with rare inclusions of sillimanite. Minor amounts of late muscovite occur within the leucocratic layers, but not in the metapelitic domains. No prograde andalusite, staurolite or muscovite has been found in the pelitic migmatites and, interestingly no K-feldspar has been found in the residual portions of the migmatites. The high proportion of sillimanite and cordierite/ garnet, the fact that metamorphic conditions probably exceed the stability of muscovite plus quartz, and the lack of K-feldspar in the most residual melanosomes suggests that a melt phase coexisted with the melanosomes. This melt phase was subsequently extracted and probably mixed with the leucosomes, since some of these leucosomes are clearly intrusive from the nearby plutons. The lack of K-feldspar and the decreasing abundance of plagioclase and quartz among the mesosomes and melanosomes (see table 1 in Jung et al. 1998) is also compatible with the view that a melt phase was extracted. Some of the plagioclase and K-feldspar may have formed cumulates in the leucosomes, since these leucosomes have small but discernable positive Eu anomalies. However, due to the absence of cumulate textures, the limited extraction distances of the leucosomes (some are only centimetres away from the melanosomes), the gradual nature of some leucosomemelanosome pairs and the evidence of progressive entrainment of xenoliths or wall rock material rather then fractional crystallisation, it is suggested that a cumulus nature of the leucosomes is, at least in part, unlikely. In addition, the amount of K-feldspar and plagioclase is way too large to be solely related to the inferred amount of these minerals in the melanosomes and consequently, most of the K-feldspar and plagioclase must be attributed to the original composition of the leucosomes. Note that a complex contamination–

accumulation modelling was performed to explain the composition of some of the larger leucosomes associated with the most residual melanosomes (Jung et al. 1998) and up to 10 wt.% of K-feldspar/plagioclase may be regarded as accumulated minerals. It is suggested that consumption of early muscovite+ quartz+plagioclase+H2O led to the formation of a ﬁrst melt, but the appearance of spinel, garnet and cordierite indicates that also a biotite-dehydration melting reaction was operative. It is therefore reasonable to assume that melt production and extraction probably occurred episodically, both at the maximum P–T conditions and during decompression of the rocks. In summary, petrographic evidence, and thermobarometric constraints, indicate that the migmatites outcropping at the Davetsaub and Nomatsaus areas experienced upper amphibolite to lower granulite facies conditions with temperatures approaching 720–750(C at 4–6 kbars (Saywer 1981; Kukla et al. 1991; Jung et al. 1998). The major and trace element abundances as well as the Nd, Sr and oxygen isotope compositions suggest that mesosome MES 2, and the melanosomes MEL 3 and MEL 4 (Fig. 2) are strongly residual rocks. The chemical and isotopic composition of the leucosomes can be best explained by a complex partial melting–mixing– accumulation process (Jung et al. 1998).

3. Geochronology of the migmatite Early age determinations on the migmatites from the Davetsaub area yielded a Rb–Sr whole-rock isochron age of 5055 Ma (Blaxland et al. 1979; Haack et al. 1982) consistent with a monazite age of 5054 Ma given by Kukla et al. (1991) for the intrusion of a late-tectonic granite in this area. Rb–Sr whole rock ages are 5238 Ma and 52115 Ma for the Donkerhoek granite in this area (Blaxland et al. 1979). U–Pb monazite ages from the migmatites give ages between 5252 Ma and 5212 Ma for composite migmatites and between 5212 Ma and 5182 Ma for monazite from neosomes (Kukla et al. 1991) providing a lower limit for the peak of metamorphism. More recently, the time of high-grade regional metamorphism and the duration of regional metamorphic events were constrained by Sm–Nd (garnet), U–Pb (monazite) and Rb–Sr (biotite) ages obtained from a composite migmatite sample (Jung & Mezger 2001). Sm–Nd garnet whole rock ages for a strongly restitic melanosome and an adjacent intrusive leucosome yield ages of 5345 Ma, 52811 Ma and 5398 Ma. These results provide substantial evidence for pre-500 Ma Pan-African regional metamorphism and melting for this segment of the orogen. Other parts of the migmatite yielded apparently younger Sm–Nd garnet whole rock ages of 4889 Ma for melanosome and 49610 Ma, 4925 Ma and 51116 Ma for the corresponding leucosomes. Garnet from one xenolith within the leucosomes yielded an age of 4972 Ma. Monazite from the leucosomes records 207Pb/235U ages of between 5361 Ma and 5291 Ma, indicating that this monazite represent incorporated residual material from the ﬁrst melting event. Monazite from the mesosome and the melanosome gives 207 Pb/235U ages of between 5231 Ma and 5311 Ma, broadly similar to the monazite ages from the leucosomes. The apparently unreacted metasedimentary rock MET 1 yielded a 207 Pb/235U monazite age of 5082 Ma, which probably indicate another later thermal event. Taken together, these ages indicate that high-grade metamorphism started at c. 535 Ma (or earlier) and was followed by thermal events between c. 520 Ma and c. 490 Ma. Rb–Sr biotite ages from the diﬀerent layers of the migmatite are c. 488 Ma, 469 Ma and 473 Ma.

TRACE ELEMENT CONSTRAINTS – A GARNET IONPROBE STUDY

209

These diﬀerent ages indicate late-stage disturbance of the Rb–Sr isotopic system on the sub-sample scale. Nevertheless, these ages are close to the youngest Sm–Nd garnet ages indicating rapid cooling rates between 13–20(C/Ma and fast uplift of this segment of the crust (Jung & Mezger 2001).

4. Results 4.1. Garnet major element composition Garnet is predominantly an almandine–pyrope–spessartine solid solution with a minor grossular component. Inclusionrich garnet from melanosome MEL 3 has the composition alm69py16sp12gr3 and the inclusion-poor garnet from the same melanosome has the composition alm71py11sp16gr2. Inclusionfree garnet from the melanosome has the composition alm70py17sp11gr2. The garnets from the leucosome and the melanosome show rather similar zoning proﬁles, with ﬂat element distributions over most of the crystal and increasing Fe/Mg at the outermost rim. There is no pronounced zoning with respect to the grossular component. Details are given in Jung et al. (1998).

4.2. Garnet trace element composition Analytical details are summarised in the Appendix (see section 8). Three texturally distinct garnets can be distinguished, an inclusion-rich (Fig. 3a) and inclusion-poor garnet (Fig. 3b) within the melanosome and an apparently inclusion-free garnet within the leucosome (Fig. 3c). The garnets have a 103 –103 CI-normalised range in REE abundances and steep LREE-depleted and HREE-enriched element patterns (Fig. 4). A negative Eu anomaly of variable magnitude is developed in the garnets. 4.2.1. Inclusion-rich garnet (melanosome). This garnet (Fig. 3a) has a core composition enriched in HREE+Y and a rim composition that is depleted in HREE+Y (Fig. 4a and b). The core–rim variation is gradual, although Yb and Er display some more enriched domains in the garnet core. Furthermore, the zoning is signiﬁcantly less pronounced for the middle REE (rim/core w0·6–1·0 for Sm, Eu, Gd), compared to the more compatible heavy REE (rim/core w0·2–0·5 for Dy–Yb) and Y (rim/core: 0·5). Furthermore, it shows an irregular or ﬂat distribution of Ti, Cr, Zr and Sc over the crystal, but a tendency for enrichment in V (and to lesser extent Na) can be observed towards the rim of the crystal (Table 1). Asymmetric distribution of Zr of the crystal seems to indicate contamination with micro-inclusions of zircon. Sr seems to be slightly enriched in the core relative to the rim, although the concentrations appear to be similar within error. The distribution of Sm follows roughly that for HREE and Y, with a peak around the core region and depletion in Sm towards the rim. On the other hand, Nd shows a broad core region with low concentrations and a rim composition with elevated concentrations (Fig. 4c). Towards the outermost rim, Nd concentrations decrease again. However, fractionation of LREE ((Sm/Nd)N ratio, where N denotes normalisation to chondrite, according to Boynton 1984) seems to be more or less constant throughout the crystal (Table 1). On the other hand, fractionation of HREE (Yb/Dy and Yb/Er ratios) is more pronounced for the core than for the rim (Fig. 5). The magnitude of the negative Eu anomaly ranges from 0·03 to 0·09 with a tendency to decrease towards the rim. 4.2.2. Inclusion-poor garnet (melanosome). This garnet has also a core composition enriched in HREE+Y and a rim composition that is depleted in HREE+Y (Fig. 6a and b). Both types of garnet have roughly similar grain sizes and

Figure 3 Photographs of garnet from the diﬀerent samples with the approximate location of the spots analysed by ion microprobe: (top) inclusion-rich garnet from melanosome MEL 3; (middle) inclusion-poor garnet from melanosome MEL 3; (bottom) inclusionfree garnet from leucosome LEU 3. Lower side of each photograph is approximately 2 mm.

equally spaced ionprobe spots. Therefore, both analysed garnets are comparable and, in contrast to the inclusion-rich garnet, the HREE+Y-enriched core composition of the inclusion-poor garnet appears to be somewhat larger than the HREE+Y-depleted rim. Core–rim variation of HREE+Y is also gradual, but in contrast to the inclusion-rich garnet, the inclusion-poor garnet has a core composition that is slightly depleted in HREE+Y and also displays HREE enrichment at the outermost rim. Ignoring the HREE-enriched outermost rim, HREE concentrations are similar to both garnet types. In

210

C. JUNG ET AL.

Sm, Eu, Gd), and for the more compatible HREE (rim/core w0·3–0·7 for Dy–Yb) and Y (rim/core: 0·3). Whereas in the inclusion-rich garnet, the core–rim distribution increases from Yb (core/rim: 0·22) and Er (core/rim: 0·32) to Dy (core/rim: 0·46), the core–rim variation in the inclusion-poor garnet decreases from Yb (core/rim: 0·68), Er (core/rim: 0·45) and Dy (core/rim: 0·28). Yttrium concentrations are similar at the outermost rim of both garnet types, but are enriched by a factor of two in the core of the inclusion-poor garnet. In contrast to the inclusion-rich garnet from the melanosome, concentration proﬁles for Ti, V, Cr, Zr and Sr show enriched core compositions and depleted rim compositions. Zr abundances are highly symmetric and cannot be explained by contamination with micro-inclusions of zircon. For Ti, there is a pronounced peak in the core region, whereas the concentration proﬁles for V, Cr, and Zr are rather ﬂat. Sr displays re-enrichment at the outermost rim. Sc shows a rather ﬂat concentration proﬁle, with lower concentrations in the core than in the rim (Table 1). The distribution of Sm and Nd follows roughly that for HREE and Y, with low values in the core region followed by enrichment of Sm and Nd at the rim. At the outermost rim, Sm and Nd concentrations are signiﬁcantly lower (Fig. 6c). Fractionation of LREE ((Sm/Nd)N) shows a core with lower ratios and a rim with higher ratios. The outermost rim shows signiﬁcantly lower ratios than both, core and rim. Fractionation of HREE (Yb/Dy and Yb/Er ratios) shows elevated ratios in the core and lower values in the rim. The outermost rim shows signiﬁcantly higher ratios than in the core and rim (Fig. 7). The magnitude of the negative Eu anomaly ranges from 0·01 to 0·06 with a tendency to decrease towards the rim. 4.2.3. Inclusion-free garnet (leucosome). Like the other two garnet types from the melanosome, the garnet from the leucosome has a core composition enriched in HREE+Y and a rim composition that is depleted in HREE+Y (Fig. 8a, b). Compared to the other garnet, the trace element-enriched core region is only 1/3 that of the total garnet. Rare Earth Element zoning is similar or less pronounced for the middle REE (rim/core w0·3–0·7 for Sm, Eu, Gd), compared to the more compatible heavy REE (rim/core w0·3–0·2 for Dy–Yb) and Y (rim/core: 0·3). The core–rim variation in HREE and Y is gradual. In contrast to the other garnets, other trace elements exhibit either pronounced core–rim depletion (Sr, Zr, Sc) or core–rim enrichment (Ti, V, Cr, Na; Table 1). Again, the core is enriched in Zr. If contamination by micro-inclusions is responsible for this selective enrichment, it would be highly fortuitous that this occurred exactly in the core region. In some cases, elements may show enrichment (Sc) or depletion (Na) at the outermost rim. The distribution of Sm and Nd follows roughly that for HREE and Y, with high values in the core region followed by depletion of Sm and Nd at the rim. At the outermost rim, Sm and Nd concentrations are slightly enriched (Fig. 8c). Fractionation of LREE ((Sm/Nd)N) shows a core with lower ratios and a rim with higher ratios. Fractionation of HREE (Yb/Dy and Yb/Er ratios) shows low ratios in the core followed by elevated values towards the rim. The outermost rim shows similar low ratios than in the core (Fig. 9). As in the inclusion-poor garnet from the melanosome, the magnitude of the negative Eu anomaly ranges from 0·01 to 0·06, with a tendency to decrease towards the rim. Figure 4 (a) Chondrite-normalised Rare Earth Element plot for core and rim for inclusion-rich garnet from melanosome MEL 3. (b) HREE variation along a rim-core–rim proﬁle. (c) Sm and Nd variation along a rim-core–rim proﬁle. Chondrite values are from Boynton (1984).

5. Discussion 5.1. Garnet Gd/Dy geothermometry

contrast to the inclusion-rich garnet from the melanosome, REE zoning is similar for the MREE (rim/core w0·3–0·8 for

Heavy Rare Earth Element variation, i.e. the Gd/Dy ratio, can be used to constrain the equilibration pressure that prevailed

TRACE ELEMENT CONSTRAINTS – A GARNET IONPROBE STUDY Table 1

211

Trace element composition (in ppm) of garnet from melanosome MEL 3 and leucosome LEU 3. For sample reference see Jung et al. (1998).

Inclusion-rich garnet MEL 3 Na Sc Ti V Cr Sr Y Zr La Ce Nd Sm Eu Gd Dy Er Yb Sm/Nd(N) Yb/Dy Yb/Er Inclusion-poor garnet MEL 3 Na Sc Ti V Cr Sr Y Zr La Ce Nd Sm Eu Gd Dy Er Yb Sm/Nd(N) Yb/Dy Yb/Er

Rim

Core

Rim

90·2 140 17·5 18·6 67·7 0·16 161 12·2 0·01 0·01 0·03 0·17 0·027 1·28 15·0 26·8 37·7

87·9 155 22·8 19·8 79·3 0·15 187 14·2 0·11 0·18 0·11 0·20 0·014 1·32 16·2 31·7 52·4

54·8 141 23·0 14·3 74·1 0·17 209 3·8 0·00 0·01 0·04 0·21 0·024 1·60 20·2 33·8 71·1

49·0 164 23·7 13·7 70·5 0·17 287 3·80 0·00 0·01 0·06 0·28 0·032 2·06 25·7 47·9 103·0

47·5 157 23·2 13·6 66·6 0·17 241 2·12 0·00 0·00 0·07 0·24 0·025 1·94 22·3 40·7 83·9

87·5 147 29·0 15·1 76·9 0·16 227 24·3 0·00 0·00 0·06 0·22 0·029 1·78 21·1 34·3 67·6

144 137 34·1 17·8 88·5 0·15 199 41·0 0·00 0·00 0·07 0·23 0·016 1·34 17·9 31·0 60·9

141 135 55·1 20·7 79·6 0·15 168 82·9 0·00 0·01 0·14 0·24 0·026 1·54 15·4 19·3 36·6

105 144 50·8 18·2 70·5 0·15 133 55·1 0·00 0·00 0·11 0·23 0·045 1·45 12·9 15·7 30·0

120 124 40·6 17·8 87·2 0·15 139 50·2 0·00 0·01 0·09 0·19 0·032 1·26 11·9 15·3 23·2

4·9 2·5 1·4

1·9 3·2 1·7

5·9 3·5 2·1

4·7 4·0 2·2

3·4 3·8 2·1

3·5 3·2 2·0

3·5 3·4 2·0

1·7 2·4 1·9

2·1 2·3 1·9

2·0 1·9 1·5

Rim

Core

Rim

89·9 124 41·9 15·9 81·5 0·19 179 16·9 0·00 0·00 0·06 0·27 0·035 2·41 20·2 33·7 69·7

95·6 88·0 161 19·2 104 0·16 182 45·0 0·00 0·01 0·18 0·71 0·084 4·93 28·0 21·6 22·7

83·6 91·5 202 21·3 105 0·18 422 55·2 0·01 0·01 0·29 1·57 0·101 10·9 64·9 44·4 43·0

90·2 98·6 164 24·3 124 0·20 606 55·8 0·01 0·01 0·23 1·20 0·034 11·1 83·2 83·0 107

96·4 97·8 613 23·4 113 0·23 544 53·2 0·01 0·01 0·12 0·66 0·044 8·41 71·6 73·7 102

94·9 104 123 23·2 114 0·24 654 54·5 0·00 0·01 0·15 0·84 0·040 9·78 86·0 90·3 124

87·4 107 101 25·2 101 0·18 534 53·4 0·01 0·02 0·21 1·00 0·035 11·2 81·5 55·8 50·5

80·9 122 217 23·3 91·2 0·18 401 56·1 0·00 0·01 0·34 1·69 0·095 11·8 60·7 39·4 36·7

71·9 149 40·8 16·7 83·6 0·23 406 13·1 0·01 0·01 0·09 0·26 0·025 3·43 44·9 102 220

4·8 3·4 2·1

4·0 0·8 1·0

5·3 0·7 1·0

5·3 1·3 1·3

5·6 1·4 1·4

5·4 1·4 1·4

4·7 0·6 0·9

5·0 0·6 0·9

2·8 4·9 2·2

during formation of the garnet (Bea et al. 1997). The studied garnets from the melanosome and leucosome show Gd and Dy zoning similar to the zoning observed for the more compatible HREE and, consequently, this type of zoning is interpreted to be the result of Rayleigh-type fractionation during growth of the garnet. It seems that Gd and Dy behave similarly and therefore Gd/Dy ratios are fairly constant over the crystal. Using the equation derived by Bea et al. (1997), average apparent equilibration pressures are 4·10·04 kbar for the inclusion-rich garnet from the melanosome, 4·40·2 kbar for the inclusion-poor garnet from the melanosome and 4·20·1 kbar for the inclusion-free garnet from the leucosome. These pressures are slightly lower than the pressures of 5·20·4 kbar and 5·60·2 kbar obtained by conventional GASP and Grt–Crd cation exchange barometry (Jung et al. 1998).

5.2. Inclusion-rich garnet (melanosome) The inclusion-rich garnet from the melanosome is part of a restitic mineral assemblage consisting of garnet, biotite, sillimanite, spinel and quartz (Jung et al. 1998). Based on distinct Sm–Nd garnet-whole rock ages and U–Pb monazite ages, the migmatite is considered to have undergone multiple high-grade episodes, including one or more metamorphic and/or anatectic events. With respect to the garnet, it is likely that the actual mineralogical composition of the melanosome records only the latter of these episodes. This suggestion is supported by a Sm–Nd garnet whole rock age of 4889 Ma, which is the youngest age among the diﬀerent parts of the migmatite (Jung & Mezger 2001). Therefore, it is suggested that the garnet trace element composition can be used to place constraints on the garnet-forming process, namely growth of garnet in a closed system with no change in garnet growth reaction. In theory,

212

C. JUNG ET AL.

Table 1 Continued. Inclusion-free garnet LEU 3 Na Sc Ti V Cr Sr Y Zr La Ce Nd Sm Eu Gd Dy Er Yb Sm/Nd(N) Yb/Dy Yb/Er

Rim

Core

Rim

99·3 113 54·2 33·8 14·9 0·12 163 7·41 0·00 0·01 0·07 0·21 0·02 1·71 18·5 16·8 27·2

113 87·6 43·7 16·9 15·2 0·10 103 7·50 0·00 0·00 0·03 0·14 0·02 1·23 11·9 14·1 27·6

140 82·7 34·1 20·3 14·7 0·12 130 8·60 0·00 0·00 0·04 0·14 0·01 1·28 13·0 17·3 40·0

80·8 144 31·0 9·6 7·83 0·17 553 63·5 0·00 0·00 0·13 0·52 0·02 6·37 63·8 76·8 135·4

84·7 109 35·3 23·1 15·2 0·16 274 8·58 0·00 0·01 0·05 0·30 0·02 2·33 20·4 31·3 73·2

71·9 116 40·5 19·3 14·0 0·13 149 3·65 0·00 0·01 0·09 0·16 0·03 1·49 10·3 14·7 25·7

111 111 62·1 24·2 13·3 0·11 99·5 10·5 0·00 0·01 0·07 0·16 0·03 1·47 11·2 13·0 15·3

95·8 139 52·9 26·4 9·80 0·11 167 13·7 0·00 0·00 0·06 0·28 0·03 2·50 22·3 21·3 34·2

3·0 1·5 1·6

4·5 2·3 2·0

3·6 3·1 2·3

4·0 2·1 1·8

6·3 3·6 2·3

1·8 2·5 1·7

2·4 1·4 1·2

4·7 1·5 1·6

Figure 5 Yb/Dy and Yb/Er variation along the same proﬁle for the inclusion-rich garnet from melanosome MEL 3.

the abundance of a trace element on the surface of a growing garnet in equilibrium with a surrounding metamorphic mineral assemblage is controlled by (i) the eﬀective bulk composition and (ii) the bulk distribution coeﬃcient (Kd) for partitioning of the element between the garnet and the matrix phases. Therefore, the zoning of elements that exhibit decreasing (Sr, Y, HREE, Sm) concentrations from core to rim can be explained in terms of equilibrium closed-system partitioning models (Hollister 1966; Cygan & Lasaga 1982; Otamendi et al. 2002). In this model, crystal diﬀusion-controlled partitioning is a Rayleigh fractionation process, whereby surface equilibrium is maintained and elemental diﬀusion is limited by the growing crystal. This process assumes mass balance in the reservoir (i.e., a closed system which in this case is the metamorphic host

rock), constant Kd values and removal of a fractionating element by mineral growth. One interesting feature is the distribution of Nd relative to Sm. Nd is more incompatible than Sm in garnet, and in the core region, at the point where Sm peaks, Nd concentrations are low followed by re-enrichment towards the rim (Fig. 4c). This feature may indicate limited volume diﬀusion of Nd during high-temperature metamorphism. Heavy Rare Earth Element fractionation, monitored by Yb/Dy and Yb/Er ratios, is continuous from core to rim and more pronounced for Yb/Dy than for Yb/Er ratios which is consistent with the diﬀerent partition coeﬃcient, Dgarnet/matrix such that DYb >DEr >DDy >DGd (Fig. 5). Otamendi et al. (2002) described garnet from an anatectic terrane in which residual garnet cores, interpreted to have grown during highgrade regional metamorphism, have high Yb/Er and Yb/Dy ratios. Garnet rims, interpreted as peritectic garnet, show inverted ratios, i.e., lower Yb/Dy and higher Yb/Er ratios. The most prominent feature of the HREE systematics was an inﬂection of the HREE ratios at the core–rim interface. The continuous decrease of HREE ratios in the inclusion-rich garnet from the melanosome may arise from equilibrium partitioning during garnet growth, but the missing inﬂection indicates that partial melting processes are not monitored by the garnet composition.

5.3. Inclusion-poor garnet (melanosome) In contrast to the other garnet from the melanosome, this garnet has much fewer inclusions (Fig. 3b). Elements such as Sr, Ti, V and Cr show core–rim depletion, whereas Sc shows core–rim enrichment (Table 1). The distribution of these elements may indicate equilibrium closed-system partitioning (Rayleigh fractionation). Generally, Heavy Rare Earth Element and Y abundances decrease from core to rim (Fig. 6a). However, in the core region, there is a slight depletion in HREE and Y relative to the adjacent rim. In addition, the outermost rim shows re-enrichment in HREE and Y (Fig. 6b). The slight depletion in the core, followed by higher values at the rim with subsequent depletion and re-enrichment at the outermost rim, suggests a multi-stage growth history. Taken

TRACE ELEMENT CONSTRAINTS – A GARNET IONPROBE STUDY

213

Figure 7 Yb/Dy and Yb/Er variation along the same proﬁle for the inclusion-poor garnet from melanosome MEL 3.

It is possible that the core–rim distribution of HREE and Y can be attributed to variable garnet growth rates during high-grade metamorphism and melting. Variable garnet growth rates are possible because it is reasonable to assume that the garnet now preserved in the melanosome has, in part, started to grow within the melanosome matrix; but subsequently equilibrated with an in-situ-derived granitic melt where additional growth of this garnet took place. Such a process may explain the increase in HREE abundances towards the rim, because there is no sink for HREE in the inferred REE-depleted leucosome. If such a reaction indeed proceeded during migmatisation, changes in the phase assemblages are unavoidable because of the distinct mineralogical composition of the host melanosome and the in situ-derived melt. Such complex distributions patterns of enrichment/depletion of HREE and Y may indicate an episode of open system behaviour. In contrast to the inclusion-rich garnet from the melanosome, the inclusion-poor garnet shows a complex HREE fractionation pattern. In the core, Yb/Dy and Yb/Er ratios are high, with Yb/Dy>Yb/Er, which is consistent with partition coeﬃcient Dgarnet/matrix:DYb >DEr >DDy. Subsequently, these ratios decrease towards the rim (Fig. 7). Towards the outermost rim, an inversion of these HREE ratios occur, with Yb/DyYb/Er. The inﬂection of these ratios between core and rim can be interpreted as a result of partial melting. Garnet cores with high Yb/Er and Yb/Dy>1 nucleated in the metamorphic environment without the presence of a melt, whereas the rims with lower Yb/Er and Yb/Dy<1 crystallised in the presence of a melt (Otamendi et al. 2002).

Figure 6 (a) Chondrite-normalised Rare Earth Element plot for core and rim for inclusion-poor garnet from melanosome MEL 3. (b) HREE variation along a rim-core–rim proﬁle. (c) Sm and Nd variation along a rim-core–rim proﬁle.

together, these features cannot reﬂect distribution of these elements following an equilibrium closed-system partitioning model.

5.4. Inclusion-free garnet (leucosome) Based on the apparent lack of inclusions, it can be suggested that this garnet is distinct to the other two garnets and probably equilibrated in the presence of a melt. The zoning of elements that exhibit either decreasing concentrations (LREE, HREE, Y, Sr, Zr, Sc) or increasing concentrations (Ti, V, Cr, Na) from core to rim can be explained in terms of a nearequilibrium closed-system partitioning model. Similar to the model that is suggested for the inclusion-rich garnet from the melanosome, crystal diﬀusion-controlled partitioning is a

214

C. JUNG ET AL.

Figure 9 Yb/Dy and Yb/Er variation along the same proﬁle for the inclusion-free garnet from leucosome LEU 3.

Figure 8 (a) Chondrite-normalised Rare Earth Element plot for core and rim for inclusion-free garnet from leucosome LEU 3. (b) HREE variation along a rim-core–rim proﬁle. (c) Sm and Nd variation along a rim-core–rim proﬁle.

Rayleigh fractionation process, whereby surface equilibrium is maintained and elemental diﬀusion is limited by the growing crystal. This process assumes mass balance in the reservoir (i.e., a closed system which in this case is the leucosome), constant Kd values and removal of a fractionating element by

mineral growth. In the initial stages of garnet growth, the relatively small core of the garnet nucleated and acted as a sink for the most compatible trace elements (i.e., HREE, Y) that show decreasing concentrations from core to rim (Figs 8 & 9). Subsequently, the volumetrically much larger and trace element-depleted rim crystallised in equilibration with the LREE-depleted leucosome. The lower Yb/Dy and Yb/Er in the core are enigmatic; one possible explanation could be that the core region represents an entrained, restitic garnet from the melanosome. However, textural evidence does not support such a suggestion. The accurate interpretation of trace element variation in garnet coexisting with a granitic liquid has important implications for trace element behaviour during partial melting processes. In general, garnet that crystallised in the presence of a melt is distinct to garnet from metamorphic environments. In contrast to garnet from strongly fractionated leucogranite (Sevigny 1993; Jung & Hellebrand 2006), the garnet from the leucosome shows a trace element distribution similar to those observed in garnet from high-grade metamorphic or migmatitic rocks. Due to the observation that the leucosome garnet is inclusion-free, incorporation of this garnet from the melanosome into the leucosome seems to be precluded. This implies that the process that formed the trace element pattern of the leucosome garnet was distinct from the liquid diﬀusion controlled process (i.e., Sevigny 1993) envisaged for leucogranitic garnets. One possible explanation could be that in hot, highly viscous, H2O-poor leucosomes from high-grade terranes, the rate of diﬀusion of trace elements is similar to the rate envisaged for high-grade metamorphic rocks, but distinct to the rate suggested for ﬂuid-dominated leucogranites.

5.5. Implications for the HREE composition of garnet-bearing leucosomes It has been suggested that the HREE composition of metamorphic and peritectic garnet can be used to make assumptions about equilibration of garnet with in situ-derived melts (Otamendi et al. 2002). Here, the inﬂection of the Yb/Dy and Yb/Er ratios at the interface between garnet core and rim was interpreted as to result from partial melting processes in which

TRACE ELEMENT CONSTRAINTS – A GARNET IONPROBE STUDY

the garnet core with its characteristic high Yb/Dy ratios nucleated in the metamorphic environment and the rim with lower Yb/Dy ratios nucleated in an environment that is characterised by HREE depletion, i.e. representing an in situ

215

partial melt. Therefore, calculated REE patterns of the inferred in situ melts can show a marked hump at Yb, using garnet core compositions and melts equilibrated with the rim that are depleted in Yb relative to Y and Er (see ﬁgure 3 of Otamendi et al. 2002 for reference). Since the inclusion-poor garnet from the melanosome and the inclusion-free garnet from the leucosome show some of these features (inverted Yb/Dy and Yb/Er ratios at the core–rim interface), it is pertinent to calculate corresponding leucosome compositions and to compare them with the leucosomes from the migmatite and with granitic melts from the Damara orogen. Figure 10a–c shows multi-element distribution pattern for equilibrium partial melts calculated from the garnet composition (core and rim) using calculated mineral-melt distribution coeﬃcients according to the approach of Hollister (1966). From Figure 10a it is obvious that, for the inclusion-rich garnet from the melanosome, calculated compositions of corresponding felsic melts either in equilibrium with the garnet core or the rim are one order of magnitude higher in HREE and Y relative to unfractionated leucosomes from the migmatite. These features may indicate that this garnet was never in equilibrium with a felsic melt similar in composition to unfractionated leucosomes from the migmatite. In the case of the inclusion-poor garnet from the melanosome, the calculated melt compositions are also enriched in HREE and Y, but the rim composition tend to approach the composition of the leucosomes from the outcrop. For the inclusion-free garnet from the leucosome, the melt composition calculated with the core composition of the garnet is markedly enriched in HREE and Y. On the other hand, the melt composition that is calculated using the garnet rim composition is depleted in HREE and Y, and is similar to other leucosome compositions from the Damara orogen. Based on the data of Otamendi et al. (2002), it was concluded that in situ-derived felsic melts were unable to equilibrate with garnet cores, but were able to equilibrate with garnet rims. Some of the features interpreted to be characteristic for peritectic garnet (i.e., inversion of Yb/Dy and Yb/Er ratios at the core–rim interface) are also observed in the migmatite garnets from the Davetsaub/Nomatsaus areas, and it seem also obvious that calculated melt compositions were in equilibrium with garnet rims but not the cores.

6. Conclusion This study demonstrates that trace element zoning in garnet porphyroblasts from a polyphase migmatite from the highgrade part of the Damara orogen (Namibia) provide a powerful tool of investigating melting processes in the crust. Trace element zoning in garnet is more pronounced than major

Figure 10 Plots of calculated abundances of Dy, Er, Yb and Y in melts, assuming equilibration with either garnet cores or garnet rims. Assumptions for the calculation of the trace element abundances were (i) garnet is the only phase that govern these trace element abundances and (ii) bulk partition coeﬃcients between garnet and matrix (melanosome or leucosome) are similar to partition coeﬃcients between garnet and melt. The latter assumption is an oversimpliﬁcation, but actual garnet/melt partition coeﬃcients are likely to be similar to garnet/ matrix partition coeﬃcients and relative values of partition coeﬃcients such as (DYb >DEr >DY >DDy). Because of the large values for D, melt fraction has a minor eﬀect on calculated trace element concentrations in the melt and, therefore, garnet/melt partitioning can be simpliﬁed to: C(i)melt/C(i)garnet =1/D(i), where C(i)melt and C(i)garnet are concentrations of an element (i) in melt and garnet, respectively. Also shown in (c) are trace element abundances in leucosomes from the Damara orogen (Jung et al. 2000, 2003).

216

C. JUNG ET AL.

element zoning, and provides information on important metamorphic processes, i.e. in situ migmatisation and metamorphic growth; information that cannot be retrieved from majorelement investigations alone. As anticipated, the garnet analysed in this study are important reservoirs for HREE and Y. The concentrations of HREE and Y in garnet cores are in most cases the result of simple Rayleigh fractionation between the growing garnet and the rock matrix. The signatures of this process are the bell-shaped element distribution patterns. However, inclusion-poor garnet from a melanosome shows a more complex trace element pattern with distinct distribution patterns, suggesting disequilibrium between garnet and an inferred in situ-derived melt. Such complex trace element distribution patterns probably mark the onset of open system processes. The preservation of such disequilibrium patterns may occur when mineral growth rates are rapid, particularly in migmatite environments where temperature overstepping may occur. Reacting metamorphic minerals and in situ-derived melts have distinct trace element compositions, and because trace element diﬀusivities in garnet are apparently slow at amphibolite facies to lower granulite facies conditions, variations in garnet trace element compositions can be used to track the metamorphic history of such rocks.

7. Acknowledgements This study was supported by a grant from the Deutsche Forschungsgemeinschaft to E. Hoﬀer (Ho 1078/12-1) and the Max-Planck Gesellschaft. The authors would like to thank G. Lugmair and Al Hofmann (MPI Mainz) for access to the ionprobe hosted by the Max Planck Institut fu¨r Chemie in Mainz. Peter Hoppe and Elmar Gro¨ner are warmly thanked for keeping the ionprobe in good shape. I. Bambach is warmly thanked for producing high quality ﬁgures. We would like to thank I. Buick and F. Bea for the instructive reviews and editor G. Stevens for smooth and patient editorial handling.

8. Appendix. Analytical techniques Garnet was analysed for trace elements (selected REE and Na, Sc, Ti, V, Cr, Sr, Y, Zr) by secondary ion mass spectrometry (SIMS) on a recently upgraded Cameca IMS-3f in Mainz. Spots were selected for ion microprobe analysis after detailed petrographic and electron microprobe study. Only optically clear domains that showed no signs of alteration or opx exsolution were analysed. Negative oxygen ions were used as primary ions (accelerating potential of 12·5 kV and 20 nA beam current). The spot size for these operating conditions was 15–20 m. For very small grains, the beam current was reduced to 10 nA, resulting in a smaller spot size (around 10 m). Positive secondary ions were extracted using an accelerating potential of 4·5 kV with a 25 eV energy window, a high-energy oﬀset of 80 V, and fully open entrance and exit slits. Each measurement consisted of a six-cycle routine, where in each cycle the species 16O, 30Si, 47Ti, 51V, 52Cr, 88Sr, 89Y, 90 Zr, 138Ba, 139La, 140Ce, 146Nd, 147Sm, 153Eu, 157Gd, 163Dy, 167 Er and 174Yb were analysed, in that order. In each cycle, the REE were measured for 30 s, Sr, Zr and Ba for 20 s, Ti, V and Y for 5 s, and the other elements for 1 s. At the beginning of each analysis, the energy distribution of 16O was measured to determine the maximum intensity and the precise location of the 10% low-energy edge of the distribution. The location of this sharp edge can be determined more precisely than the location of relatively broad maximum intensity. Given that Umax-edge is +20 V, the applied high-energy oﬀset is about 100 V from the location of the 10% value of this steep ﬂank.

In this way, diﬀerences in ion energy as a result of charge build-up from one sample to the next do not aﬀect the energy range of the ions being analysed (Zinner & Crozaz 1986). Subsequently, peak centres were determined for 30Si, 47Ti, 89Y and 163Dy by scanning the peak in 20 steps across a 1·5 wide B-ﬁeld. The neighbouring masses (Cr and V on Ti, Sr and Zr on Y, and all REE on Dy) were then adjusted to these new peak centres. From one measurement to the next, however, the peak shift was rarely signiﬁcant (<50 ppm). For all silicates, 30 Si (3·1% isotopic abundance) is used as a reference mass, as the SiO2 concentration of standards and samples is known from electron microprobe analysis. For each cycle, mass to 30Si ratios were determined after correction for time-dependence of count rates, detector deadtime (20 ms) and background [10 3 c.p.s. (counts per second)]. The average of these ratios was used to calculate the element concentration, multiplying the measured ratios by a constant factor. These so-called sensitivity factors were determined for each element on the wellstudied standard glasses KL2–G, ML3B-G, StHs6/80–G, BM90/21–G and ATHO-G (Jochum et al. 2000). For this purpose, a diﬀerent measurement routine that determines the mass spectrum between 133 and 191 was adopted (Zinner & Crozaz 1986), referred to below as the ‘long routine’. This approach is necessary to obtain accurate sensitivity factors for the REE. Although the applied energy ﬁltering technique (Shimizu et al. 1978) eliminates the eﬀect of most molecular interferences, it is well known that element monoxides can produce signiﬁcant interferences, particularly on REE (e.g. PrO on Gd). These have to be corrected. The intensity of an interfering oxide is a function of three parameters: the absolute concentration element of the interfering oxide (i.e. shape of the REE pattern), its isotopic abundance, and the oxide/element ratio of the interfering species. In contrast to the short routine described above, the long routine provides an internally consistent method to determine the corrected element/Si ratios and the oxide/element ratio. A detailed description of this iterative data reduction procedure was presented by Zinner & Crozaz (1986). Owing to the low LREE concentrations in garnet, the short measurement routine was preferred over the long routine, as the poor counting statistics at the extremely low counting rates caused an unacceptable propagating error and long duration time for a single analysis. Therefore, the short routine used the MO + /M + values obtained by the long routine. Furthermore, six oxides were found to produce signiﬁcant interferences: 137BaO + interferes with 153Eu + , 141PrO + with 157 Gd + , 147SmO + with 163Dy + , 151EuO + with 167Er + , and both 158GdO + and 158DyO + with 174Yb + , and their MO + to M + ratios used for the corrections are 0·046, 0·13, 0·06, 0·05, 0·08 and 0·07, respectively (with errors <10%). As all elements of these oxides are free of interferences, they were measured directly. For the concentrations of Eu, Gd, Dy and Er in a typical LREE-depleted garnet, these corrections were always <2% of the measured element/Si ratio. For Yb, the correction was 11%. The well-studied glass GOR132–G (Jochum et al. 2000) was used as an external standard, as its LREE-depleted pattern is broadly similar to that of the measured garnet (Table 1). The overall accuracy is between 7% and 19% for the REE and better than 11% for all other elements (95% conﬁdence level). To distinguish ‘real’ counts from background noise at extremely low counting rates, separate background measurements with long counting times of between 10 and 30 min were carried out. All reported analyses are corrected for the background, which lies around 10 3 (c.p.s.) on average. The background concentration for each element that is based on such a count rate is controlled primarily by the intensity of 30 Si, which was (1·0–1·3)105 c.p.s. Deﬁning a conservative

TRACE ELEMENT CONSTRAINTS – A GARNET IONPROBE STUDY

detection limit at the six-fold background level, this corresponds to values for the REE that range from 0·6 ng/g (La) to 4 ng/g (Sm).

9. References Bea, F., Montero, P., Garuti, G. & Zacharini, F. 1997. Pressuredependence of rare-earth element distribution in amphibolite- and granulite-grade garnets. A LA-ICP-MS study. Geostandards Newsletters 21, 253–70. Blaxland, A., Gohn, E., Haack, U. & Hoﬀer, E. 1979. Rb/Sr ages of late-tectonic granites in the Damara orogen, Southwest Africa/ Namibia. Neues Jahrbuch fu¨r Mineralogie Monatshefte 11, 498–508. Boynton, W. V. 1984. Geochemistry of rare earth elements: Meterorite studies. In Henderson, P. (ed.) Rare earth element geochemistry, 63–114. Amsterdam: Elsevier. Brown, M. 1983. The petrogenesis of some migmatites from the Presqu’ile de Rhuys, Southern Brittany, France. In Atherton, M. P. & Gribble, C. D. (eds) Migmatites, Melting and Metamorphism 174–200. Nantwich: Shiva Press. Chernoﬀ, C. B. & Carlson, W. D. 1999. Trace element zoning as a record of chemical disequilibrium during garnet growth. Geology 27, 555–8. Cygan, R. & Lasaga, A. C. 1982. Crystal growth and the formation of chemical zoning in garnet. Contributions to Mineralogy and Petrology 79, 187–200. Essene, E. J. 1989. The current status of thermobarometry in metamorphic rocks. In Daly, J. S., Cliﬀ, R. A. & Yardley, B. W. D. (eds) Evolution of metamorphic belts. Geological Society, London, Special Publication 43, 1–44. Bath, UK: The Geological Society Publishing House. Gray, D. R., Foster, D. A., Goscombe, B., Passchier, C. W. & Trouw, R. A. J. 2006. Ar40/Ar39 thermochronology of the Pan-African Damara Orogen, Namibia, with implications for tectonothermal and geodynamic evolution. Precambrian Research 150, 49–72. Haack, U., Hoefs, J. & Gohn, E. 1982. Constraints on the origin of Damaran granites by Rb/Sr and 18O data. Contributions to Mineralogy and Petrology 79, 279–89. Hartmann, O., Hoﬀer, E. & Haack, U. 1983. Regional metamorphism in the Damara orogen: Interaction of crustal motion and heat transfer. In Miller, R. McG. (ed.) Evolution of the Damara orogen. Special Publication of the Geological Society, South Africa 11, 233–41. Hickmott, D. D., Shimizu, N., Spear, F. S. & Selverstone, J. 1987. Trace element zoning in a metamorphic garnet. Geology 15, 573–6. Hickmott, D. D. & Shimizu, N. 1990. Trace element zoning in garnet from the Kwoiek area, British Columbia: disequilibrium partitioning during garnet growth? Contributions to Mineralogy and Petrology 104, 619–30. Hickmott, D. D. & Spear, F. S. 1992. Major and trace element zoning in metamorphic garnets from western Massachusetts. Journal of Petrology 33, 965–1005. Hoernes, S. & Hoﬀer, E. 1979. Equilibrium relations of prograde metamorphic mineral assemblages – a stable isotope study of rocks of the Damara orogen. Contributions to Mineralogy and Petrology 68, 377–89. Hollister, L. S. 1966. Garnet zoning: an interpretation based on the Rayleigh fractionation model. Science 154, 1647–51. Jochum, K. P., Dingwell, D. B., Rocholl, A., Stoll, B., Hofmann, A. W., Becker, S., Besmehn, A., Dietze, H.-J., Dulski, P., Erzinger, J., Hellebrand, E., Hoppe, P., Horn, I., Janssens, K., Jenner, G. A., Klein, M., McDonough, W. F., Maetz, M., Mezger, K., Mu¨nker, C., Nikogosian, I. K., Pickhardt, C., Raczek, I., Rhede, D., Seufert, H. M., Simakin, S. G., Sobolev, A. V., Spettel, B., Straub, S., Vincze, L., Wallianos, A., Weckwerth, G., Weyer, S., Wolf, D. & Zimmer, D. 2000. The preparation and preliminary characterisation of eight geological MPI-DING standard reference glasses for in-situ microanalysis. Geostandards Newsletter 24, 87–133. Jung, S. 2005. Garnet and staurolite as potential geochronometers in low to medium grade metamorphic rocks – U/Pb and Pb/Pb garnet and staurolite ages from the Southern Damara orogen (Namibia). Geophysical Research Abstracts 7, EGU05–A-06266. Jung, S., Mezger, K., Masberg, P, Hoﬀer, E. & Hoernes, S. 1998. Petrology of an intrusion-related high-grade migmatite: implications for partial melting of metasedimentary rocks and leucosome-forming processes. Journal of Metamorphic Geology 16, 425–45.

217

Jung, S., Hoernes, S., Masberg, P. & Hoﬀer, E. 1999. The petrogenesis of some migmatites and granites (Central Damara orogen, Namibia): Evidence for disequilibrium melting, wall-rock contamination and crystal fractionation. Journal of Petrology 40, 1241–69. Jung, S., Hoernes, S. & Mezger, K. 2000. Geochronology and petrology of stromatic and nebulitic migmatites from the Proterozoic Damara Belt – importance of episodic ﬂuid-present disequilibrium melting and consequences for granite petrology. Lithos 51, 153–79. Jung, S., Hoernes, S. & Mezger, K. 2003. Petrology of basementdominated terranes: II. Contrasting isotopic (Sr, Nd, Pb, and O) signatures of basement-derived granites and constraints on the source region of granite (Damara orogen, Namibia) Chemical Geology 199, 1–28. Jung, S. & Hellebrand, E. 2006. Trace element fractionation during high-grade metamorphism and crustal melting – constraints from ion microprobe data of metapelitic, migmatitic and igneous garnets and implications for Sm–Nd garnet geochronology. Lithos 87, 193–213. Jung, S. & Mezger, K. 2001. Geochronology in migmatites – An Sm–Nd, U–Pb and Rb–Sr study from the Proterozoic Damara belt (Namibia) and implications for polyphase development of migmatites in high-grade terranes. Journal of Metamorphic Geology 19, 77–9. Jung, S. & Mezger, K. 2003. Petrology of basement-dominated terranes: I. Regionalmetamorphic P–T–t path from U–Pb monazite and Sm–Nd garnet geochronology (Central Damara orogen, Namibia). Chemical Geology 198, 223–47. Kukla, C. 1993. Strontium isotope heterogeneities in amphibolite facies, banded metasediments – a case study from the late Proterozoic Kuiseb Formation of the Southern Damara orogen, Central Namibia. Memoirs of the Geological Survey Namibia 15, 1–139. Kukla, C., Kramm, U., Kukla, P. A. & Okrusch, M. 1991. U–Pb monazite data relating to metamorphism and granite intrusion in the northwestern Khomas Trough, Damara Orogen, central Namibia. Communications of the Geological Survey of Namibia 7, 49–54. Masberg, H. P., Hoﬀer, E. & Hoernes, S. 1992. Microfabrics indicating granulite-facies metamorphism in the low-pressure central Damara Orogen, Namibia. Precambrian Research 55, 243–57. Miller, R. McG. 1983. The Pan-African Damara Orogen of South West Africa/Namibia: In Miller, R. McG. (ed.) Evolution of the Damara orogen. Special Publication of the Geological Society, South Africa 11, 431–515. Nieberding, F. 1976. Die Grenze der zentralen Granitzone su¨dwestlich Otjimbingwe (Karibib Distrikt, Su¨dwestafrika): Intrusionsverband, Tektonik, Petrographie. Go¨ttinger Arbeiten zur Geologie und Pala¨ontologie 19, 1–78. Otamendi, J. E., de la Rosa, J. D., Patino Douce, A. E. & Castro, A. 2002. Rayleigh fractionation of heavy rare earths and yttrium during metamorphic garnet growth. Geology 30, 159–62. Puhan, D. 1983. Temperature and pressure of metamorphism in the Central Damara orogen. In Miller, R. McG. (ed.) Evolution of the Damara orogen. Special Publication of the Geological Society, South Africa 11, 219–23. Saywer, E. W. 1981. Damaran structural and metamorphic geology in an area southeast of Walvisbay, SWA/Namibia. Memoirs of the Geological Survey, South West Africa/Namibia 7, 1–94. Sevigny, J. H. 1993. Monazite controlled Sm/Nd fractionation in leucogranites: An ion microprobe study of garnet phenocrysts. Geochimica et Cosmochimica Acta 57, 4095–102. Shimizu, N., Semet, M. P. & Alle`gre, J. C. 1978. Geochemical applications of quantitative ion microprobe analysis. Geochimica et Cosmochimica Acta 42, 1321–34. Spear, F. S., Hickmott, D., Crowley, P. & Hodges, K. V. 1984. P–T paths from garnet zoning: a new technique for deciphering tectonic processes in crystalline terrains. Geology 12, 87–90. Spear, F. S. & Florence, F. P. 1992. Thermobarometry in granulites: Pitfalls and new approaches. Precambrian Research 55, 209–41. Spear, F. S. & Kohn, M. J. 1996. Trace element zoning in garnet as a monitor of crustal melting. Geology 24, 1099–102. Spear, F. S. & Selverstone, J. 1983. Quantitative P–T paths from zones minerals: theory and tectonic implications. Contributions to Mineralogy and Petrology 83, 348–57. Thompson, A. B. 1982. Dehydration melting of pelitic rocks and the generation of H2O-undersaturated granitic liquids. American Journal of Science 282, 1567–95.

218

C. JUNG ET AL.

Tracy, R. J. 1982. Compositional zoning and inclusions in metamorphic minerals. In Ferry, J. M. (ed.) Characterization of metamorphism through mineral assemblages in metamorphic rocks. Mineralogical Society of America Reviews in Mineralogy 10, 355–97. Vielzeuf, D. 1983. The spinel and quartz associations in high grade xenoliths from Tallante (S.E. Spain) and their potential use in geothermometry and barometry. Contributions to Mineralogy and Petrology 82, 301–11.

Ward, R., Stevens, G. & Kisters, A. 2008. Fluid and deformation induced partial melting and melt volumes in low-temperature granulite-facies metasediments, Damara Belt, Namibia. Lithos 105, 253–71. Zinner, E. & Crozaz, C. 1986. A method for the quantitative measurement of rare earth elements in the ion probe. International Journal of Mass Spectrometry and Ion Processes 69, 17–38.

MS received 5 December 2007. Accepted for publication 10 November 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 219–233, 2010 (for 2009)

Crustal melt granites and migmatites along the Himalaya: melt source, segregation, transport and granite emplacement mechanisms M. P. Searle, J. M. Cottle*, M. J. Streule and D. J. Waters Department of Earth Sciences, Oxford University, Parks Road, Oxford OX1 3PR, UK *Current address: Department of Earth Sciences, University of California, Santa Barbara, CA 93106, USA ABSTRACT: India–Asia collision resulted in crustal thickening and shortening, metamorphism and partial melting along the 2200 km-long Himalayan range. In the core of the Greater Himalaya, widespread in situ partial melting in sillimanite+K-feldspar gneisses resulted in formation of migmatites and Ms+Bt+Grt+TurCrdSil leucogranites, mainly by muscovite dehydration melting. Melting occurred at shallow depths (4–6 kbar; 15–20 km depth) in the middle crust, but not in the lower crust. 87Sr/86Sr ratios of leucogranites are very high (0·74–0·79) and heterogeneous, indicating a 100% crustal protolith. Melts were sourced from fertile muscovite-bearing pelites and quartzo-feldspathic gneisses of the Neo-Proterozoic Haimanta–Cheka Formations. Melting was induced through a combination of thermal relaxation due to crustal thickening and from high internal heat production rates within the Proterozoic source rocks in the middle crust. Himalayan granites have highly radiogenic Pb isotopes and extremely high uranium concentrations. Little or no heat was derived either from the mantle or from shear heating along thrust faults. Mid-crustal melting triggered southward ductile extrusion (channel ﬂow) of a mid-crustal layer bounded by a crustal-scale thrust fault and shear zone (Main Central Thrust; MCT) along the base, and a low-angle ductile shear zone and normal fault (South Tibetan Detachment; STD) along the top. Multi-system thermochronology (U–Pb, Sm–Nd, 40Ar–39Ar and ﬁssion track dating) show that partial melting spanned ~24–15 Ma and triggered mid-crustal ﬂow between the simultaneously active shear zones of the MCT and STD. Granite melting was restricted in both time (Early Miocene) and space (middle crust) along the entire length of the Himalaya. Melts were channelled up via hydraulic fracturing into sheeted sill complexes from the underthrust Indian plate source beneath southern Tibet, and intruded for up to 100 km parallel to the foliation in the host sillimanite gneisses. Crystallisation of the leucogranites was immediately followed by rapid exhumation, cooling and enhanced erosion during the Early–Middle Miocene. KEY WORDS: Channel ﬂow, crustal melting, granite emplacement, Himalaya, leucogranite, migmatite, shear zones.

The collision of the Indian plate passive continental margin with the southern active continental margin of Asia (Karakoram in the west; Lhasa block in Tibet) is an ongoing process that started in the Palaeocene. Final marine sedimentation along the Indus–Yarlung Tsangpo suture zone was Early Eocene (49–50·5 Ma) and crustal thickening and shortening by folding and thrusting propagated southward across the north Indian plate margin (Searle et al. 1988, 1990, 1997; Rowley 1996; Zhu et al. 2005). Since the collision, India has continued to converge and penetrate north into Asia, resulting in double normal thickness continental crust both along the Indian plate margin (Himalaya) and along the south Asian margin (Karakoram–Lhasa Block). The Himalayan upper crust (Tethyan Himalaya) is composed of folded and thrust Phanerozoic (Late Precambrian to Eocene) sedimentary rocks bounded along the north by the Indus–Tsangpo suture zone and along the south by the South Tibetan detachment (STD), a low-angle, north-dipping normal fault (Burg 1983; Hodges 2000; Cottle et al. 2007). South of this the Greater Himalayan Sequence (GHS) is composed of regional Barrovian facies metamorphic rocks, migmatites and leucogranites, bounded along the south by a 2–4 km-thick zone

of inverted metamorphic isograds (from sillimanite–kyanite down to biotite–chlorite) with a brittle thrust fault along the base, the Main Central thrust (MCT) zone (Searle et al. 2008; Figs. 1, 2). The Lesser Himalaya to the south is composed of underthrust Indian plate rocks including Proterozoic basement and thin Palaeozoic cover sedimentary rocks. The deeper, unexposed lower crust beneath the Himalaya is thought to be underthrust granulite facies Indian shield rocks (Searle et al. 2006; Jackson et al. 2008). Since the Moho steepens northwards from ca. 40 km depth beneath the Himalayan foreland to ca. 80 km beneath southern Tibet (Schulte-Pelkum et al. 2005), or even 90 km beneath the Karakoram–west Tibet (Rai et al. 2006), these Precambrian granulites may have undergone a phase transition to eclogite facies rocks during the late Tertiary–present day (Searle et al. 2006). Deep seismic proﬁling, combined with broadband earthquake and magnetotelluric data across southern Tibet suggest that a high-conductivity layer at 15–20 km-depth reﬂect partial melting in the middle crust of southern Tibet today (Nelson et al. 1996; Wei et al. 2001). ‘Bright spots’ of high electrical conductivity probably reﬂect pockets of leucogranite magmas forming today at similar P–T conditions and depth as the

2009 The Royal Society of Edinburgh. doi:10.1017/S175569100901617X

220

M. P. SEARLE ET AL.

Figure 1 Geological map of the Himalaya showing U–(Th)–Pb ages along the orogen. See text and Godin et al. (2006) for sources of data.

Figure 2 Cross-section across the Everest Himalaya, based on surface geological mapping combined with deep crust seismic constraints from INDEPTH (Nelson et al. 1996).

Miocene leucogranites along the Himalaya (Gaillard et al. 2004). Seismic reﬂectors bounding this mid-crustal zone of partial melting can be traced to the ca. 20–16 Ma MCT and

STD shear zones along the Greater Himalaya (Hauck et al. 1998; Searle et al. 2006). Seismic tomographic studies suggest that southern Tibet and the Himalaya are underlain by cold,

HIMALAYAN CRUSTAL MELT GRANITES AND MIGMATITES

strong upper mantle attached to the underthrusting Indian shield (Tilmann et al. 2003). Following the India–Asia collision, crustal thickening and shortening (by folding and thrusting) along the northern margin of India resulted in increased pressure and temperatures. Progressive metamorphism evolved with time and in space from initial UHP coesite–eclogite-grade aﬀecting only the leading margin of the Indian plate, to regional kyanite- and sillimanite-grade Barrovian metamorphism aﬀecting most of the GHS and North Himalayan domes. During the Early Miocene increasing temperature and decreasing pressure resulted in decompression melting within the upper part of the middle crust of the Himalaya. One of the most spectacular results of the Himalayan collision was the formation of crustal melt leucogranites from a widespread migmatitic partial melt zone in the middle crust. Many of the highest peaks of the Greater Himalaya are composed of these anatectic granites (e.g. Shivling, Bhagirathi, Thalay Sagar, Manaslu, Makalu, Shisha Pangma, Kangchenjunga, the base of the Everest– Nuptse–Lhotse massif, Chomolhari and Masang Kang). Since the major shear zones, faults and fabrics across the Himalaya dip to the north and many of the major river drainage access routes run north–south, it is possible to map out a three-dimensional view of the entire middle and upper crust. Thus the Himalaya is a unique orogen where it is possible to trace granitic melts from their mid-crustal source to their present high structural level. This paper reviews the ﬁeld relationships, metamorphic history and chronology of metamorphism, melting and deformation along the Himalaya. The protolith source of leucogranite melts is discussed, and the partial melting process and melt reactions. These data are used to propose a model for the generation and emplacement of Himalayan leucogranites from source to high structural level, and discuss the role of partial melting and leucogranite formation in the Channel Flow model (Beaumont et al. 2001; Grujic et al. 2002; Searle et al. 2003, 2006; Searle & Szulc 2005; Jessup et al. 2006; Law et al. 2006).

1. Review of metamorphic history of the Himalaya The thermal history of the Himalaya involves four ‘stages’ during part of a 50 m.y. continuum: (1) eclogite metamorphism during initial crustal subduction of the leading edge of India to UHP depths (27·5 kbar; >100 km depth; 720–770(C) at 46·4 Ma (U–Pb, zircon, allanite; Parrish et al. 2006); (2) crustal thickening resulted in peak kyanite grade metamorphism (550–680(C; 9–11 kbar) at ca. 37–30 Ma (U–Pb, monazite and Sm–Nd garnet ages; Walker et al. 1999; Vance & Harris 1999); followed by (3) widespread sillimanite grade metamorphism (620–770(C; 4·5–7 kbar) accompanied by partial melting and leucogranite formation at ca. 23–16 Ma (Noble & Searle 1995; Walker et al. 1999; Simpson et al. 2000; Viskupic & Hodges 2001). Burial and thickening, followed by heating, decompression, partial melting and rapid exhumation and cooling resulted in clockwise P–T–t paths (e.g. Hubbard 1989; Searle et al. 1999a, b; Hodges 2000; Walker et al. 2001; Jamieson et al. 2004; Fig. 3). Himalayan eclogite metamorphism is only known so far for certain from Kaghan, North Pakistan and Tso Morari, Ladakh, although it could have occurred elsewhere along the leading margin of the Indian plate and still remain unexposed. In North Pakistan the UHP metamorphism and the medium P–T metamorphic events were probably synchronous and juxtaposed by later thrusting (Treloar et al. 2003). The kyanite and sillimanite grade metamorphic events are common along the entire length of the GHS. A fourth Himalayan metamorphic event, characterised

221

Figure 3 Pressure–temperature diagram showing key equilibria and P–T paths relevant to melting in the GHS. Filled circles are P–T conditions determined for migmatites in the Khumbu Valley, Everest region, from Searle et al. (2003). Bold arrowed curve is the P–T path determined for the migmatitic core of the GHS in the Zanskar Himalaya, from Searle et al. (1999a). Curves for signiﬁcant metamorphic and melting reactions in muscovite- and biotite-bearing schists are based on the calculated phase diagram for ‘average pelite’ in White et al. (2001).

by low-P high-T metamorphism, metasomatism and generation of cordierite-bearing leucogranites, has only been recorded in the NW Himalayan syntaxis at Nanga Parbat (Whittington et al. 1998, 1999; Zeitler et al. 2001a, b; Crowley et al. 2009) and the NE Himalayan syntaxis at Namche Barwa (Booth et al. 2004). Himalayan leucogranites were generated during the Early Miocene along the entire 2200 km length of the Greater Himalaya. Crustal melting occurred at relatively shallow depths of ca. 15–20 km depth, and granites were generated from a widespread partial melting migmatite zone in the middle crust. Many Himalayan leucogranites remain more-orless in situ within the sillimanite–K-feldspar migmatite zone and have been exposed only because of underthrusting of successive thrust slices beneath, and subsequent hanging-wall uplift and erosion. Where melt migration has occurred, kilometre-scale, foliation-parallel sill complexes within the upper part of the GHS have transported magma horizontally or along very shallow north-dipping melt channels. No leucogranites cut up across the South Tibetan Detachment (e.g. Murphy & Harrison 1999; Searle & Godin 2003). U–(Th)–Pb ages of leucogranites have been used to constrain timing of motion along the ductile shear zone and the STD low-angle normal fault according to whether the leucogranites are pre-, syn- or post-kinematic with relation to the ductile fabric and brittle fault (e.g.: Searle et al. 1997, 2003, 2006). In Nanga Parbat, Proterozoic granulite basement gneisses were rehydrated to amphibolite before undergoing Oligocene– Miocene metamorphism and at least four phases of melting. The youngest of these was Pleistocene and associated with a vigorous hydrothermal system (Zeitler et al. 2001a, b). Cordierite–K-feldspar–quartz pods and veins (Butler et al. 1997; Whittington et al. 1998, 1999; Whittington & Treloar 2002; Crowley et al. 2009) intrude along vertical extensional fractures in the core region of the Nanga Parbat massif. Cordierite occurs as coronas around Al and Fe–Mg rich

222

M. P. SEARLE ET AL.

Figure 4 (a) South face of Lingtren (6749 m) showing a 3–4 km-thick sheeted sill complex of horizontallylayered leucogranite with a massive sill at the top enclosing rafts of sillimanite gneiss. (b) Layered leucogranites interlayered with GHS gneisses along footwall of the STD in the Rongbuk valley, north of Everest, south Tibet. (c) North face of Ama Dablam (6828 m) showing leucogranite sheets or thick sills with horizontal dykes bent to the north along the footwall of the STD. (d) Gyachung Kang (7922 m), west of Everest, showing the two low-angle normal faults, the upper Qomolangma detachment and the lower Lhotse detachment.

material which are probably small restitic enclaves. The coronas merge into coarse-grained cordierite in the presence of melt. The development of cordierite-bearing corona assemblages should reﬂect the ﬁnal interaction between melt and solid at a time no earlier than that recorded by the crystallisation age of accessory phases in the leucosome. P–T conditions of cordierite coronas from the core of Nanga Parbat are 3·7–4·2 kbar at 700–720(C (Crowley et al. 2009). U–Th–Pb ages from monazites and xenotimes in leucogranites give ages as young as 0·76–0·71 Ma (Bowring et al. 2004; Crowley et al. 2005, 2009). Exhumation rates could be as high as 1·5–2 cm/yr, and 13–15 km of overburden has been eroded from the summit region of Nanga Parbat in less than one million years. In the Eastern Himalayan syntaxis at Namche Barwa, U–Pb SHRIMP zircon ages show the standard 25–18 Ma Himalayan melt events, but also a younger period of 10–3 Ma in the core of the massif (Booth et al. 2004), similar to the Nanga Parbat syntaxis.

2. Field relationships of Himalayan granites Himalayan leucogranites are almost entirely found within the upper part of the Greater Himalayan mid-crustal slab, within the sillimanite+K-feldspar migmatite zone, or intruded along giant sill complexes beneath the STD. Similar ﬁeld relationships are seen along the entire length of the Himalaya from Zanskar (e.g. Searle et al. 1999a; Walker et al. 1999; Dezes et al. 1999) through Garhwal (e.g. Scaillet et al. 1990, 1995; Searle et al. 1999b) to Nepal (e.g. Searle & Godin 2003),

Sikkim (e.g. Searle & Szulc 2005) and Bhutan (e.g. Grujic et al. 2002). Although early studies suggested that the Manaslu leucogranite was an exception by intruding across the STD into the overlying base of the Tethyan sediments (e.g. LeFort 1975, 1981; Guillot et al. 1995; Harrison et al. 1999), Searle & Godin (2003) showed that the granite was emplaced into high-grade metamorphic rocks beneath the STD, and does not intrude across the STD. The metamorphism around the Manaslu granite is clearly continuous with the regional GHS metamorphism well away from the granite (Searle & Godin 2003), and not contact metamorphism around the granite (Guillot et al. 1995). In the Everest area almost all leucogranites are layer-parallel sills of thicknesses varying from <1 m to 3–4 km (Searle 1999a, b, 2003, 2007; Searle et al. 2003, 2006; Jessup et al. 2006, 2008; Cottle et al. 2007, 2009). The sills can be traced in the ﬁeld for up to 40 km horizontally across strike, within the upper part of the GHS from the upper Khumbu glacier in Nepal and along the Rongbuk glacier in Tibet (Searle et al. 2003, 2006; Jessup et al. 2006, 2008; Cottle et al. 2007, 2009). Massive horizontal sill complexes make up the upper part of the GHS, with the larger sills reaching 3–4 km thick. The massive granite sheet exposed around the base of the Everest massif is continuous west to Lingtren (Fig. 4a) Pumori, Gyachung Kang and Cho Oyu, east to the granites surrounding the upper Kangshung valley and the base of Makalu, and south to the ballooning Nuptse granite (Searle 1999a, b, 2003, 2007; Jessup et al. 2006, 2008). At Rongbuk, early folded and later foliationparallel sills decrease in abundance up-section to the Lhotse

HIMALAYAN CRUSTAL MELT GRANITES AND MIGMATITES

detachment (Fig. 4b), above which a prominent band of calc-silicate truncates the rocks beneath. The brittle Qomolangma detachment above the calc-silicate band and the Everest series pelites then dips north beneath the Tibetan plateau some 35–40 km north of Everest (Cottle et al. 2007). Vertical dykes feed magma up from lower sills to higher sills. The highest dykes have been bent around to the north along the footwall of the Lhotse detachment (e.g. on the north face of Ama Dablam; Fig. 4c). West of Mount Everest on Gyachung Kang (Fig. 4d) there are clearly two lowangle detachments, the ﬁrst of which separates the Cambro-Ordovician sedimentary rocks on the summits from the Everest series greenschist–lower amphibolite facies metapelites and calcschists (Qomolangma detachment) and the second is the lower ductile Lhotse detachment which separates the Everest series above from sillimanite gneisses with abundant leucogranites beneath (Searle 1999a, b, 2003; Searle et al. 2003, 2006; Jessup et al. 2006, 2008). At deeper structural levels the melting zone, represented by the migmatites, shows layered or stromatic migmatites with in situ melts sweating out of quartzo-feldspathic or K-feldspar augen gneisses. The Manaslu leucogranite in central Nepal is one of the larger Himalayan granites, being ca. 5 km thick (Fig. 5a). The upper margin of the granite is an abrupt fault-bounded contact that corresponds to the Phu detachment, the upper brittle STD normal fault (Searle and Godin 2003) that places Cambrian sedimentary rocks directly above the granite (Fig. 5b). The leucogranite contains rafts of the migmatitic gneiss from which it was apparently derived (Fig. 5c). Tourmaline is a common maﬁc phase within the Manaslu granite that also contains abundant muscovite, some biotite and garnet (Fig. 5d). Common textures include tourmaline+quartz ‘ghosts’ that form circular pods of metasomatic late mineral growth associated with boron-rich ﬂuids (Fig. 5e). The stromatic migmatites and early leucogranite melts are cut by later dykes of leucogranite fed by later melting episodes deeper in the crust (Fig. 5f). Most Himalayan leucogranite bodies are made up of a variety of two-micatourmalinegarnet leucogranite. Cordierite is uncommon in most Himalayan leucogranites, but it is widespread in the Makalu area, east of Everest, as well as in the younger Pliocene–Recent Nanga Parbat migmatitic leucosomes and very young leucogranite melts (Whittington et al. 1998; Bowring et al. 2004; Crowley et al. 2005, 2009). On Makalu the youngest intrusive phase is a 3 km-thick sill-like body of cordierite granite (Fig. 6a, b) that is fed by a 10 m-wide vertical dyke that cuts across earlier fabrics and older two mica tourmaline leucogranite sills (Fig. 6c). Cordierite occurs as large pale green crystals that are interpreted as a low-pressure (<5 kbar) peritectic phase formed as a product of the biotite dehydration reaction (Fig. 6d). In the Makalu area multiple dykes record at least six episodes of crustal melting in batches (Fig. 6e, f).

3. Review of the U–Th–Pb ages of Himalayan granites The ﬁrst attempts to date the crystallisation age of Himalayan leucogranites utilised the 87Sr/86Sr method (e.g. Deniel et al. 1987). However, the highly variable 87Sr/86Sr ratios made calculating ages by this method problematic. Most subsequent age determinations have therefore relied on the U(–Th)–Pb technique. This method is not without its problems however. Because anatexis occurred under low-T conditions, and the resulting melts are highly peraluminous, xenocrystic accessory phases are common. Most have inherited zircon; some have

223

inherited monazite (Copeland et al. 1988); whilst inherited xenotime, although rare, is known (Viskupic & Hodges 2001). An additional problem is that monazite, the mineral of choice when dating Himalayan leucogranites, is usually reversely discordant in U–Pb space. This behaviour is interpreted to reﬂect an initial U–Th disequilibrium caused by the incorporation of excess 230Th during crystallisation, leading to an excess of 206Pb (Scha¨rer 1984). The presence of excess 206Pb results in the 206Pb/238U ages for monazite in this study being older than the 207Pb/235U and 208Pb/232Th ages by as much as 50% (Cottle et al. 2009). U–Pb dates can be corrected for the U–Th disequilibrium by estimating the degree of U–Th fractionation between mineral and melt (Scha¨rer 1984; Parrish 1990), but given that the Th–Pb system and the 207Pb/235U ages are unaﬀected by this disequilibrium the 207Pb/235U or 208 Pb/232Th dates are generally taken as the most reliable estimates of the ages of the grains measured. Most U(–Th)–Pb ages for the melts in the central Himalaya (Fig. 1) are Early–Middle Miocene, ranging from 24–15 Ma (Harrison et al. 1995; Hodges et al. 1996; Coleman 1998; Searle et al. 1997, 1999a, b; Godin et al. 2001; Daniel et al. 2003; Harris et al. 2004; Cottle et al. 2009) to 13–12 Ma (Edwards & Harrison 1997; Wu et al. 1998; Zhang et al. 2004). However, evidence for leucosome melt production during the Oligocene (33–23 Ma) also exists (Coleman 1998; Thimm et al. 1999; Godin et al. 2001). In the Everest transect the oldest record of in situ partial melting occurs in the Namche orthogneiss at w25–26 Ma (Viskupic & Hodges 2001). Immediately following initial melting, a major phase of pre- to syn-kinematic melt mobilisation occurred between w22 and 21 Ma (Simpson et al. 2000; Viskupic & Hodges 2001; Searle et al. 2003; Viskupic et al. 2005). The timing of this earliest melt mobilisation is broadly synchronous with emplacement of many of the large granite sheets present at the highest structural level in the Everest region GHS, such as the Everest/Makalu granite (Scha¨rer 1984; Simpson et al. 2000). Early foliation parallel granite sheets have U–Pb monazite ages spanning 24–18 Ma (Scha¨rer 1984; Viskupic & Hodges 2001; Searle et al. 2003; Viskupic et al. 2005; Cottle et al. 2009), whereas later sets of undeformed dikes cross-cut ductile fabrics and have U(–Th)–Pb ages of w18–16 Ma (Hodges et al. 1998; Murphy & Harrison 1999; Simpson et al. 2000; Viskupic & Hodges 2001; Searle et al. 2003; Viskupic et al. 2005).

4. Melt reactions During progressive metamorphism, water-saturated melting may occur if a hydrous ﬂuid phase is present when rocks cross the wet granite solidus. At mid-crustal pressure, appropriate simpliﬁed reactions might be: Ms+Qtz+Pl+Kfs+H2O=Melt Ms+Qtz+Pl+H2O=Sil+Melt Unless there is an external supply of aqueous ﬂuid, however, the amount of melt generated by these vapour- (or ﬂuid-) present processes will be trivial, and signiﬁcant melt generation will require access to the water locked up in hydrous minerals such as micas or amphiboles. These reactions are referred to as vapour-absent or dehydration melting reactions, and are incongruent melting reactions yielding solid peritectic products. Vapour-absent dehydration melting reactions generally have positive P–T slopes, with the two important consequences that, ﬁrst, they can be intersected by rocks on a decompression path, and secondly, that in principle the melts generated are able to rise a considerable distance in the crust before reaching their solidus. The ﬁrst such reaction encountered by a typical

224

M. P. SEARLE ET AL.

Figure 5 (a) Massive cliﬀs, approximately 3000 metres high, of pale leucogranite exposed on the south face of Manaslu (8163 m). (b) Upper contact of the Manaslu leucogranite truncated by the north-dipping STD low-angle normal fault; north of the Larkye-la, Nepal–Tibet border. (c) Toumaline leucogranite enclosing rafts of migmatitic gneiss, Bimthang, Manaslu west. (d) Tourmaline+garnet leucogranite, Bimthang, south of Manaslu. (e) Metasomatic ‘ghosts’ of tourmaline+quartz schorl within leucogranite, Manaslu west. (f) Leucogranite dyke cross-cutting gneissic fabrics, Jangle Karkar, south of Manaslu.

metapelite at mid-crustal depths is likely to be the muscovite dehydration melting reaction: Ms+Qtz+Pl=Kfs+Als (Sil)+Melt Experimental studies in the pressure range 6–10 kbar suggest that this curve is located between about 710(C and 790(C at 8 kbar (Peto¨ 1976; Patin˜o-Douce & Harris 1998). Its calculated position, in the middle of this range, is shown in Figure 3. The amount of melt generated is typically of the order of ten volume percent, but depends primarily on the amount of muscovite present. Large migmatite terranes represent the in situ melt region. During muscovite melting, if the restite and

liquid remain in equilibrium, melting will continue until the solid muscovite reactant phase is exhausted. The biotite dehydration melting reaction: Bt+PlAls (Sil)+Qtz=KfsCrdGrt+Melt occurs at higher temperature and lower pressure than muscovite melting. Le Breton & Thompson (1988) located the beginning of biotite dehydration melting between 760(C and 800(C at 10 kbar, although signiﬁcant volumes of melt were not produced until temperatures reached 850(C. Pelitic source rocks will melt at lower temperatures (ca. 750(C; 5 kbar) whereas greywacke source rocks lacking sillimanite require

HIMALAYAN CRUSTAL MELT GRANITES AND MIGMATITES

225

Figure 6 (a), (b) Massive cordierite-bearing leucogranite on the south face of Makalu (8485 m) showing abundant xenoliths along the base of the uppermost large sill. (c) Vertical feeder dyke to the uppermost cordierite leucogranite sill on the south face of Makalu. (d) Large green cordierite crystals enclosed in Makalu leucogranite, Barun glacier, Nepal. (e), (f) Six phases of cross-cutting leucogranite dykes, south face of Makalu, Barun glacier.

higher temperatures of about 825(C at 5 kbar (Vielzeuf & Holloway 1988; Stevens et al. 1997). Biotite dehydration melting results in formation of peritectic garnet and/or cordierite in metapelites and peritectic orthopyroxene in metagreywackes. For typical metapelite bulk compositions,

cordierite is the stable phase at pressures less than about 5 kbar, whereas garnet is stable at higher pressure. Himalayan metamorphism records clockwise P–T–t paths with sillimanite grade overprinting earlier kyanite grade conditions. Harris & Massey (1994) concluded that most

226

M. P. SEARLE ET AL.

Himalayan granites were decompression melts formed on the exhumation part of the P–T path in the sillimanite ﬁeld. However, the uncommon presence of kyanite-bearing migmatites, e.g. in the Marsyandi valley, Nepal (Coleman 1998), in Bhutan (Daniel et al. 2003) and in Sikkim, is evidence that melting probably started earlier (23–16 Ma from garnet Sm–Nd ages in Harris et al. 2004), and at higher pressure (8–12 kbar) in some parts of the Himalaya. P–T phase relations and trace element modelling suggest that most Himalayan leucogranites formed by vapour-absent incongruent melting of muscovite in the temperature range 650–750(C at pressures between 9 and 4 kbar (Scaillet et al. 1990, 1996; Harris & Massey 1994; Searle et al. 1999b; Prince et al. 2001). A ﬂux of aqueous ﬂuid into the melting zone will augment the volume of granitic melt formed. There is geochemical evidence that melting was not typically an equilibrium process (Harris & Inger 1992), and that extracted melt entrained a proportion of restitic material. Incomplete mixing between diﬀerent melt batches resulted in isotopic heterogeneity in crystallised leucogranite bodies (e.g. Deniel et al. 1987; Scaillet et al. 1996). These crustal melt reactions require no heat or chemical input from the mantle. The achievement of high temperatures in the mid-crust would, however, be assisted by high internal heat production in the source region. In a ductile deforming channel such as the Early Miocene GHS, internal heat production could include both radiogenic and strain heating. In the following sections it is suggested that this is the case.

5. Source rocks Himalayan leucogranites are strongly peraluminous, characterised by the presence of muscovite and tourmaline, with biotite and garnet also present in lesser amounts. Major element compositions are fairly homogeneous, but trace elements are highly variable and depend on the nature of the source. Sr, Nd and Pb isotopes are all indicative of a metasedimentary source (Deniel et al. 1987; Harris et al. 1995, Guillot & LeFort 1995). 87Sr/86Sr ratios of leucogranites are very high (0·74–0·79) and heterogeneous, suggesting a 100% crustal protolith. Harris & Massey (1994) made a detailed Sr isotope study of the Langtang leucogranites and concluded that the protolith was not the sillimanite migmatites into which the melts have been emplaced, but rather the kyanite-bearing meta-pelites structurally lower down the GHS section. They further implied that the granites were not in situ melts but had migrated some distance (>10 km) from their source. However, in most Himalayan proﬁles mapped by the present authors it is possible to physically trace the migmatite leucosomes coalescing into giant foliation-parallel sill networks, which feed larger plutons (e.g. Searle et al. 1997, 1999b). It is perfectly possible that the leucogranite bodies were tapping diﬀerent migmatites at depth rather than ones immediately adjacent (Harris & Massey 1994) but the heterogeneous Sr isotopes probably reﬂect a wide range of protoliths with each batch of melt produced. The leucogranite sill complexes are almost always within the sillimanite grade gneisses and melt migration was more horizontal, not vertical. The larger plutons are not actually intrusive into higher structural levels, but are more like subhorizontal ballooning sills (e.g. Nuptse leucogranite; Searle et al. 2003, Searle 2003, 2007). This is the case in Zanskar (e.g. Noble & Searle 1995), Shivling, Garhwal (Searle et al. 1993, 1999b), Manaslu (Searle & Godin 2003), Shisha Pangma (Searle et al. 1997), and the Everest region (Searle 1999a, b, 2003; Searle et al. 2003, 2006; Jessup et al. 2008).

Isotopic signatures can be used to determine the nature of granite protoliths, provided that the liquid is in isotopic equilibrium with the source. Trace element modelling of partial melting is only useful if equilibrium between the melt and the solid phases has been achieved (Harris & Inger 1992; Harris et al. 1995). Himalayan leucogranites are, however, rarely in isotopic equilibrium as a result of complex sedimentary protoliths and multiple batch melting (Deniel et al. 1987). Guillot & LeFort (1995) proposed a bimodal origin of Himalayan leucogranites with two-mica leucogranites (87Sr–86Sr ratio <0·752) derived from a meta-greywacke source and tourmaline leucogranites (87Sr–86Sr ratio >0·752) derived from a metapelitic source. However, in the ﬁeld there is a complete range of granite composition with variable amounts of tourmaline, muscovite, biotite and garnet. It seems more likely that progressive batch melting tapped diﬀerent source rocks ranging from black shales to meta-greywackes within the NeoProterozoic Haimanta Formation. Cambro-Ordovician augen gneisses (Formation 3; LeFort 1975, 1981) also commonly show in situ melt textures with tourmalinegarnet bearing leucosomes.

6. Heat source for crustal melting It has become increasingly clear that peak sillimanite grade metamorphism, crustal melting and ductile shear along the MCT along the base of the GHS and ductile shear along the STD at the top of the GHS were synchronous, and these metamorphic, magmatic and structural processes must be genetically linked. The greatest problem has been to explain why peak temperatures and granite melt generation in thickened Himalayan crust were at a relatively shallow depth (15–20 km depth; 4–6 kbar), and why melting does not occur at greater depths. P–T–t data across the GHS along the Everest proﬁle show that high temperatures (>620(C) were maintained for ca. 15 million years (from 32 to 17 Ma) along the top of the GHS and that approximately 45–50 km width (20 km structural thickness) of the GHS was approximately isothermal in the sillimanite grade (Searle et al. 2003, 2006; Jessup et al. 2006, 2008). This shallow heating requires an unusually steep geothermal gradient, and crustal thickening alone cannot account for the thermal proﬁle. Restoration of the GHS shows that the source for granite melts was the Neo-Proterozoic Haimanta–Cheka Formation. A combination of high internal radiogenic heat production from the melt source rocks and thermal relaxation after crustal thickening is proposed. Average heat production in the migmatites is a factor of two more than heat production in the schists (Harris & Massey 1994). Shear heating along the MCT (England et al. 1992; Harrison et al. 1998) cannot explain the heat distribution in the GHS because maximum temperatures are a long way up-section from the MCT. No granite melts are present along the exposed MCT anywhere along the Himalaya. No mantle heat input is present, so the only possible extra heat source for restricted shallow-level melting is internal heat production by highly radioactive sedimentary source rocks. Gariepy et al. (1985) noted that Himalayan granites have highly radiogenic Pb isotopic compositions which imply sources enriched in U and Th. Uranium concentrations from Himalayan granites are some of the highest found anywhere in granitic rocks (Pinet & Jaupart 1987).

7. Melt segregation – migmatites The core of the GHS shows a vast migmatite terrane that stretches from the ductile shear zone along the STD at the top to the region above the zone of inverted metamorphic

HIMALAYAN CRUSTAL MELT GRANITES AND MIGMATITES

227

Figure 7 (a) Typical migmatite textures showing stromatic migmatite with melts generated along the foliation planes, south of Manaslu in the Marsyandi valley, Nepal. (b) Early leucogranites boudinaged in the foliation plane and subsequently folded, south of Manaslu, Nepal. (c) Melt extraction textures where melt is channelled out of leucosome into dykes that mobilise and cross-cut the migmatite fabric, south of Manaslu, Nepal. (d) Melt mobilisation from leucosome to channelled ﬂow along dykes, south of Manaslu. (e) Early layered leucogranite sills (1) in gneisses, cut by late granite melts mobilised into sill-dyke networks (2), south of Manaslu, Nepal. (f) Late tourmaline leucogranite melts breaking up host gneisses into xenoliths, south of Manaslu, Marysandi valley, Nepal.

isograds along the MCT at the base. The zone of partial melting above the sillimanite–K-feldspar isograd can reach a maximum structural thickness of 15–20 km along the Zanskar proﬁle in the western Himalaya (Searle & Rex 1989; Searle et al. 1992, 1999a, b), and 20 km along the Everest proﬁle in Nepal (Searle et al. 2003, 2006; Jessup et al. 2006) and the Kangchenjunga proﬁle in Sikkim (Searle & Szulc 2005). Although the protoliths diﬀer (from Proterozoic in the south to Cambro-Ordovician in the north), metamorphic grade, P–T conditions and degrees of partial melting are similar across the GHS. In situ crustal melting forming

migmatites is common in GHS quartzo-feldspathic gneisses, pelitic gneisses and K-feldspar augen gneisses. From ﬁeld evidence it seems likely that all these lithologies constituted source rocks for granite melts. Melt extraction pathways can be mapped out using leucosome networks. Layered stromatic migmatites are the most common textures, with melt segregations ﬂattened in the foliation plane (Fig. 7a), similar to compaction bands (e.g. Brown 2007). K-feldspar augen gneisses commonly show in situ melting, for example in the Namche orthogneiss (Viskupic & Hodges 2001; Searle et al. 2003). Kinematic

228

M. P. SEARLE ET AL.

indicators (rotated K-feldspar augen, C–S fabrics, etc.) and quantitative strain data from the Everest proﬁle show that although south-directed simple shear is dominant across the GHS, a signiﬁcant component of pure shear is apparent (Law et al. 2004; Jessup et al. 2006). This general shear is compatible with ﬂow parallel to the extension direction in horizontal sheeted leucogranite sills. U–Pb dating of monazite shows that migmatite leucosomes have Early Miocene ages, similar to the higher level leucogranites (Noble & Searle 1995; Walker et al. 1999; Dezes et al. 1999), supporting a genetic relationship between migmatite and granite, despite the diﬀerence in Sr and O isotopes (Harris & Massey 1994). Pervasive melt migration can initially occur by porous melt ﬂow through hot, viscous crust (Weinberg & Searle 1999). This is like an intermediate stage between a migmatite terrane and a sheeted sill complex. Cross-cutting relationships show batch melting of leucogranite with early formed granites folded in with the schistosity (Fig. 7b). Early leucogranite sills commonly show boudinage fabrics, indicating extension at right angles to the ﬂattening direction. Leucosomes are randomly distributed in the core of the GHS where exposed, but begin to coalesce into discrete veins and melt channels that tend to follow planes of anisotropy. As soon as leucosome in situ melts become interconnected, the melt extraction pathway is formed and the melt may start to ﬂow (Fig. 7c, d). High-temperature igneous textures show that melt extracted from the migmatite feeds dykes may cross-cut the same migmatite fabric (Fig. 7e). Thus very short time intervals between melt extraction, dyking and crystallisation are involved in each batch of melt. Later foliation-parallel sheeted sill complexes feed magma, generated from the migmatite zone, horizontally along the foliation planes of anisotropy. Later dykes cross-cut the migmatite fabric and contain ﬂoating xenoliths of migmatitic gneiss within (Fig. 7f). Melt ﬂows down pressure gradients by the easiest path, feeding magma to higher level sills. In the Himalaya these ﬂow pathways gently up-dip from north to south along north-dipping foliation planes.

8. Melt transport; granite emplacement mechanisms Several Himalayan leucogranites have been mapped out in 3-D and studied in detail. These include the Shivling–Bhagirathi granites in north India (Scaillet et al. 1990, 1995; Searle et al. 1993), the Manaslu leucogranite in west Nepal (Guillot et al. 1993, 1995; Searle & Godin 2003), the Shisha Pangma leucogranite in south Tibet (Searle et al. 1997) and the Everest– Makalu leucogranites in east Nepal–south Tibet (Searle 1999a, b, 2003; Searle et al. 2003, 2006; Jessup et al. 2006). Spectacular sheeted sill networks of leucogranite melts can be seen in many 3-D cliﬀ sections along the Himalaya. Sills propagate by hydraulic fracturing, forcing cracks apart to accommodate magma injected from the migmatite zone. Composite intrusions show that multiple batches of magma are injected into the same zone. Occasional dykes connect sills, enabling magma to be channelled up structural section. Magma batches may be separated by short time intervals. Some outcrops show foliation-parallel sills concordant with the metamorphic fabric in the host sillimanite gneiss, which feed magma into a dyke that clearly crosscuts the same metamorphic fabric. In the Himalaya there is almost no evidence of diapiric ascent of granite magma. Despite some earlier studies describing diapiric intrusions of the Manaslu leucogranite up across the STD into Tethyan sediments (LeFort 1975, 1981; Colchen

et al. 1986; Harrison et al. 1999), subsequent work has shown that the granite was intrusive into high-grade marbles, augen gneiss and pelite of the GHS and that the STD wraps around the top of the leucogranite (Searle & Godin 2003). All Himalayan leucogranites are within the GHS, beneath the STD passive roof fault. In the deeper structural levels magma ﬂow is almost entirely sub-horizontal along foliation-parallel sill complexes. At higher structural levels the sills amalgamate into larger sills and, as they approach the surface, they balloon up into inﬂated sills. This ballooning sill structure is typiﬁed by the Nuptse sill in the Everest region (Searle 1999a, b, 2003) and by the Shisha Pangma leucogranite in south Tibet (Searle et al. 1997). Even the largest Himalayan leucogranites such as the Manaslu (ca. 5 km thick) leucogranite are tabular sill-like bodies dipping gently north and truncated along the top by the STD ductile shear zone and low-angle normal fault (Searle & Godin 2003; Fig. 5b). Similarly, the Kangchenjunga leucogranite is a series of composite sills totalling approximately 12–14 km structural thickness (Searle & Szulc 2005), dipping gently north and truncated by the STD along the top. In addition to ﬁeld observations, the petrological and isotopic heterogeneity of leucogranites, and a thin upper thermal aureole, has been used as evidence for episodic emplacement of magmas to form larger leucogranite bodies. Thermal modelling of this emplacement (Annen et al. 2006) implies a 20–60 ka repeat time for the emplacement of 20–60 m-thick sills which form the larger 5 km-thick Himalayan leucogranite bodies (e.g. Manaslu, Fig. 5). To provide such a volume of melt, a highly fertile source must be heated for a prolonged time, producing low viscosity (104·5PaS) magma that is emplaced via shear assisted melt extraction (Scaillet & Searle 2006).

9. Channel Flow model The Himalayan Channel Flow model (Fig. 8) describes a protracted ﬂow of a weak, viscous, partially molten layer of middle crust between relatively rigid, but still deformable bounding upper and lower crust slabs (e.g. Beaumont et al. 2001, 2004; Grujic et al. 2002; Godin et al. 2006; Grujic 2006). Ductile extrusion of high-grade metamorphic rocks, and leucogranite melts between a coeval normal sense shear zone above (STD) and a thrust-sense shear zone below (MCT) allowed southward extrusion of the GHS. The resultant geometry of the metamorphic isograds shows an inverted and condensed P–T proﬁle above the Main Central Thrust along the base of the extruding slab (Searle & Rex 1989; Searle et al. 2008) and a right-way-up P–T proﬁle beneath the South Tibetan Detachment along the top of the extruding slab (Searle et al. 2003, 2006; Jessup et al. 2006). Extrusion was driven ultimately by the crustal thickness and topographic variation between the Tibetan plateau hinterland (70–80 km thick crust; ca. 5 km elevation) and the Indian foreland (35– 40 km thick crust, 0–1 km elevation). Numerous geological studies across the GHS have shown that almost all the data supports the Channel Flow model for the GHS during the Early Miocene. These data include P–T proﬁles (e.g. Searle et al. 2003, 2006; Jessup et al. 2006), 3-D distribution of partial melts and granites in the middle crust (e.g. Searle 1999a, b; Searle et al. 1999a, b, 2006; Searle & Szulc 2005), and quantitative data on strain, deformation temperatures and vorticity of ﬂow (e.g. Law et al. 2004; Jessup et al. 2006). Beaumont et al. (2001) used a thermal-mechanical numerical model to show that channel ﬂow and ductile extrusion were dynamically linked to the eﬀects of surface erosion focused along the Greater Himalaya at the extrusion front. As pointed out by Klemperer (2006) this model is consistent with

HIMALAYAN CRUSTAL MELT GRANITES AND MIGMATITES

229

Figure 8 Himalayan channel ﬂow model (after Searle et al. 2006; Cottle et al. 2009), a model that satisﬁes all geological and geophysical requirements in the Greater Himalaya. Inset top left shows the STD proﬁle along Dzakaa chu, north of Everest, South Tibet. Inset bottom right shows the central part of the channel in the Kangshung valley east of Everest, south Tibet. Giant blocks or rafters of gneisses with early leucogranite sills are completely enclosed in Miocene leucogranite.

all the geological data . . . ‘inevitably, since the model was designed a posteriori to ﬁt the observations’. Searle et al. (1988) and Searle & Rex (1989) ﬁrst proposed that inverted, condensed metamorphic isograds from kyanite

to biotite grade along the MCT ductile shear zone along the base of the GHS channel were linked to the right-way-up isograds along the footwall of the STD along the top of the GHS. Kinematic indicators and strain requires the GHS to be

230

M. P. SEARLE ET AL.

extruding south relative to the stable seismogenic upper crust above (Tethyan Himalaya), and the subducting Indian shield and Lesser Himalayan cover rocks beneath. Later models based on the Bhutan Himalaya (Grujic et al. 1996, 2002) included the folded isograds in their wedge extrusion or layer extrusion models. Deep crustal seismic proﬁling now reveals that the partially molten mid-crustal layer imaged beneath southern Tibet can match the geologically constrained crosssections across the GHS in Nepal (Nelson et al. 1996; Searle et al. 2003, 2006). Extensive geophysical measurements in Tibet also imply that temperatures and rheologies require some form of crustal ﬂow. Seismic velocities beneath southern Tibet are so low that the crust has to be ﬂuid-rich, and predicted temperatures are so high that these ﬂuids are likely to be partial melts. Low elastic thickness, low topographic relief and shallow cut-oﬀ of crustal seismicity all show that ﬂuids weaken the middle crust (Klemperer 2006). The ‘bright spots’ as imaged by magnetotelluric studies beneath the Lhasa block (Wei et al. 2001; Unsworth et al. 2005), are at the same structural depth as the depth of formation of Himalayan leucogranites (Searle et al. 2003, 2006; Galliard et al. 2004) and in a similar down-dip structural position in the middle crust. Unlike the middle crust, the lower crust beneath the Himalaya and southern Tibet is relatively rigid and not melting, suggesting that it is composed of dry granulite facies metamorphic rocks of the underthrust lower Indian crust. Because crustal thickness reaches 75–80 km beneath southern Tibet (Nelson et al. 1996) or even up to 90 km beneath the Karakoram and western Tibet (Rai et al. 2006), the lower crust beneath these regions must be in high-pressure granulite (dry) or eclogite (wet) metamorphic facies today (Searle et al. 2003, 2006). Brown & Solar (1998) and Brown (2006, 2007) proposed that melt loss in the middle crust might leave a dry granulite residue behind in the lower crust. The lower crust in the Himalaya is inherited Precambrian granulite facies rocks, which are anhydrous and strong and can also explain the deep crustal seismicity beneath southern Tibet (Priestley et al. 2008).

10. Conclusions The unique three-dimensional exposures around Himalayan leucogranites allow a comprehensive view of their internal and external structure to be mapped out. Extensive study of fabrics in the host gneisses, migmatites, leucogranite sills, dykes and larger bodies combined with detailed U–(Th)–Pb dating of peak metamorphic, migmatite leucosome and granite melting have allowed the fourth dimension, time, to be incorporated into models of melt generation and emplacement. Along the 2200 km length of the Himalaya, the larger leucogranite bodies always occur at similar structural horizons within the upper part of the GHS, beneath the STD low-angle normal fault. The STD forms a passive roof (stretching) fault (Searle et al. 2003; Law et al. 2004), beneath which channel ﬂow and ductile extrusion of the GHS middle crust occurred. Himalayan crustal melting occurred along the upper part of the middle crust (4–6 kbar; 15–20 km depth), but not in the lower crust. Sr, Nd and O isotopes indicate pure crustal melting with no input from the mantle. Melts were sourced from fertile muscovite-bearing pelites and quartzo-feldspathic gneisses of the Neo-Proterozoic Haimanta–Cheka Formations. The primary heat source was probably from high internal heat production rates within the Proterozoic source rocks in the middle crust. A vast in situ migmatite terrane generated melts from the melting of a heterogeneous variety of protolith rocks. Interconnected leucosome melts aggregated to force magma

into layer-parallel sills and a few cross-cutting dykes. Early Miocene leucogranites (24–17 Ma) are largely concordant with the foliation, whereas later ones (16–12 Ma) may cross-cut the ductile fabrics. All leucogranites are cut by the uppermost brittle low-angle normal fault, the Qomolangma detachment. Field observations, structural sections and geophysical evidence all provide support for the Channel Flow model, a description of the evolution of the Himalayan–south Tibetan crust during the Miocene. The Himalayan channel ﬂow model is restricted both in time (Early–Middle Miocene) and space (Himalayan–south Tibet middle crust). It is not applicable to the earlier, Eocene subduction-related UHP coesite eclogite metamorphism, or to the Oligocene HP kyanite metamorphism. Whether channel ﬂow is active today in the middle crust of southernmost Tibet–northernmost Himalaya is still open to debate (Nelson et al. 1996; Hodges 2000; Searle et al. 2006). The present model (Fig. 8) shows that the Greater Himalaya may represent the Miocene carapace of an actively extruding midcrustal channel operating at depth beneath southern Tibet today.

11. Acknowledgements This work was carried out using NERC grant NER/K/S/2000/ 00951 to MPS and PhD grants to JMC (New Zealand TEC scholarship) and MJS (NERC). We are grateful to Randall Parrish, Steve Noble, Bruno Scaillet, Laurent Godin, Rick Law, Micah Jessup and the late Doug Nelson for discussions. We also thank Peter Treloar and Alan Whittington for detailed and insightful reviews, and Alex Kisters for editorial and scientiﬁc comments.

12. References Annen, C., Scaillet, B. & Sparks, R. S. J. 2006. Thermal constraints on the emplacement rate of a large intrusive complex: The Manaslu leucogranite, Nepal Himalaya. Journal of Petrology 47, 71–95. Beaumont, C., Jamieson, R. A., Nguyen, M. H. & Lee, B. 2001. Himalayan tectonics explained by extrusion of a low-viscosity crustal channel coupled to focused surface denudation. Nature 414, 738–42. Beaumont, C., Jamieson, R. A., Nguyen, M. H. & Medvedev, S. 2004. Crustal channel ﬂows: 1. Numerical models with applications to the tectonics of the Himalayan–Tibetan orogen. Journal of Geophysical Research – Solid Earth 109, B06406. Booth, A. L., Zeitler, P. K., Kidd, W. S. F., Wooden, J., Liu, Y. P., Idleman, B., Hren, M. & Chamberlain, C. P. 2004. U–Pb zircon constraints on the tectonic evolution of southeastern Tibet, Namche Barwa area. American Journal of Science 304, 889–929. Bowring, S. A., Searle, M. P., Waters, D. J., Hodges, K. V., Schmitz, M. & Crowley, J. 2004. ID–TIMS geochronology of ca. 1 Ma leucogranites from the core of Nanga Parbat. American Geophysical Union Fall Meeting San Francisco, V22E-08. Brown, M. 2006. Melt extraction from the lower continental crust of orogens: the ﬁeld evidence. In Brown, M. & Rushmer, T. (eds) Evolution and Diﬀerentiation of Continental Crust, 332–84. Cambridge: Cambridge University Press. Brown, M. 2007. Crustal melting and melt extraction, ascent and emplacement in orogens: mechanisms and consequences. Journal of the Geological Society London 164, 709–30. Brown, M. & Solar, G. S. 1998. Shear zone systems and melts: feedback relations and self-organization in orogenic belts. Journal of Structural Geology 20, 211–27. Burg, J. P. 1983. Geological Map of South Tibet. Beijing: Ministry of Geology. Butler, R. W. H., Harris, N. B. W. & Whittington, A. G. 1997. Interactions between deformation, magmatism and hydrothermal activity during active crustal thickening: a ﬁeld example from Nanga Parbat, Pakistan Himalayas. Mineralogical Magazine 61, 37–51. Colchen, M., LeFort, P. & Pecher, A. 1986. Annapurna–Manaslu– Ganesh Himal. Carte geologique au 1:200 000 scale. Paris: CNRS. 136 pp. Coleman, M. E. 1998. U–Pb constraints on Oligocene–Miocene deformation and anatexis within the central Himalaya, Marsyandi Valley, Nepal. American Journal of Science 298, 553–71.

HIMALAYAN CRUSTAL MELT GRANITES AND MIGMATITES Copeland, P., Parrish, R. R. & Harrison, T. M. 1988. Identiﬁcation of inherited radiogenic Pb in monazite and its implications for U–Pb systematics. Nature 333, 760–3. Cottle, J. M., Jessup, M. J., Newell, D. L., Searle, M. P., Law, R. D. & Horstwood, M. S. A. 2007. Structural insight into the ductile evolution of an orogen-scale detachment: the South Tibetan Detachment System, Dzakaa Chu section, Eastern Himalaya. Journal of Structural Geology 29, 1781–97. doi:10.1016/j.jsg. 2007.08.007. Cottle, J. M., Searle, M. P., Horstwood, M. S. A. & Waters, D. J. 2009. Timing of mid-crustal metamorphism, melting and deformation in the Mount Everest region of Southern Tibet revealed by U(–Th)–Pb geochronology. Journal of Geology 117(6), 643–64. doi: 10.1086/605994 Crowley, J. L., Bowring, S. A. & Searle, M. P. 2005. U–Th–Pb systematics of monazite, xenotime, and zircon from Pleistocene leucogranites at Nanga Parbat (Pakistan Himalaya). Geochimica et Cosmochimica Acta 69, 8. Crowley, J. L., Waters, D. J., Searle, M. P. & Bowring, S. A. 2009. Pleistocene melting and rapid exhumation of the Nanga Parbat massif, Pakistan: Age and P–T conditions of accessory mineral growth in migmatite and leucosome. Earth and Planetary Science Letters 288, 408–20. doi: 10.1016/j.epsl.2009.09.044 Daniel, C. G., Hollister, L. S., Parrish, R. R. & Grujic, D. 2003. Exhumation of the Main Central Thrust from lower crustal depths, Eastern Bhutan Himalaya. Journal of Metamorphic Geology 21, 317–34. Deniel, C., Vidal, P., Fernandez, A., Le Fort, P. & Peucat, J. J. 1987. Isotopic study of the Manaslu granite (Himalaya, Nepal); inference on the age and source of Himalayan leucogranites. Contributions to Mineralogy and Petrology 96, 78–92. Dezes, P., Vannay, J.-C., Steck, A., Bussy, F. & Cosca, M. 1999. Synorogenic extension: Quantitative constraints on the age and displacement of the Zanskar shear zone (northwest Himalaya). Geological Society of America Bulletin 111, 364–74. Edwards, M. A. & Harrison, T. M. 1997. When did the roof collapse? Late Miocene north–south extension in the high Himalaya revealed by Th–Pb monazite dating of the Khula Kangri Granite. Geology 25, 543–6. England, P., LeFort, P., Molnar, P. & Peˆcher, A. 1992. Heat sources for Tertiary metamorphism and anatexis in the Annapurna– Manaslu region central Nepal. Journal of Geophysical Research 97, 2107–28. Gaillard, F., Scaillet, B. & Pichavant, M. 2004. Evidence for presentday leucogranite pluton growth in Tibet. Geology 32, 801–4. Gariepy, C., Allegre, C. & Xu, R. H. 1985. Pb-isotope geochemistry of granitoids from the Himalaya–Tibet collision zone: implications for crustal evolution. Earth and Planetary Science Letters 74, 220–34. Godin, L., Parrish, R. R., Brown, R. L. & Hodges, K. V. 2001. Crustal thickening leading to exhumation of the Himalayan metamorphic core of central Nepal; insight from U–Pb geochronology and 40 Ar/39Ar thermochronology. Tectonics 20, 729–47. Godin, L., Grujic, D., Law R. D. & Searle, M. P. 2006. Channel ﬂow, ductile extrusion and exhumation in continental collision zones: an introduction. In Law, R. D., Searle, M. P. & Godin, L. (eds) Channel Flow, Ductile Extrusion and Exhumation in Continental Collision Zones. Geological Society, London, Special Publications 268, 1–23. Grujic, D. 2006. Channel Flow and continental collision tectonics. In Law, R. D., Searle, M. P. & Godin, L. (eds) Channel Flow, Ductile Extrusion and Exhumation in Continental Collision Zones. Geological Society, London, Special Publications 268, 25–37. Grujic, D., Casey, M., Davidson, C., Hollister, L. S., Kundig, R., Pavlis, T. & Schmid, S. 1996. Ductile extrusion of the Higher Himalayan crystalline in Bhutan: Evidence from quartz microfabrics. Tectonophysics 260, 21–43. Grujic, D., Hollister, L. S. & Parrish, R. R. 2002. Himalayan metamorphic sequence as an orogenic channel: insight from Bhutan. Earth and Planetary Science Letters 198, 177–91. Guillot, S., Peˆcher, A., Rochette, P. & LeFort, P. 1993. The emplacement of the Manaslu granite of Central Nepal: ﬁeld and magnetic susceptibility constraints. In Treloar, P. J. & Searle, M. P. (eds) Himalayan Tectonics. Geological Society, London, Special Publications 74, 413–28. Guillot, S., LeFort, P., Peˆcher, A., Barman, M. R. & Aprahamian, J. 1995. Contact metamorphism and depth of emplacement of the Manaslu granite (central Nepal): implications for Himalayan orogenesis. Tectonophysics 241, 99–119. Guillot, S. & LeFort, P. 1995. Geochemical constraints on the bimodal origin of High Himalayan leucogranites. Lithos 35, 221–34.

231

Harris, N. B. W., Ayres, M. & Massey, J. 1994. Geochemistry of granitic melts produced during the incongruent melting of muscovite; implications for the extraction of Himalayan leucogranite magmas. Journal of Geophysical Research B Solid Earth and Planets 100, 15767–77. Harris, N. B. W., Ayres, M. W. & Massey, J. 1995. Geochemistry of granitic melts produced during the incongruent melting of muscovite: implications for the extraction of Himalayan leucogranite magmas. Journal of Geophysical Research 100, 15767–77. Harris, N. B. W., Caddick, M., Kosler, J., Goswami, S., Vance, D. & Tindle, A. G. 2004. The pressure–temperature–time path of migmatites from the Sikkim Himalaya. Journal of Metamorphic Geology 22, 249–64. Harris, N. B. W. & Inger, S. 1992. Trace-Element Modeling of Pelite-Derived Granites. Contributions to Mineralogy and Petrology 110, 46–56. Harris, N. B. W. & Massey, J. A. 1994. Decompression and anatexis of Himalayan metapelites. Tectonics 13, 1537–46. Harrison, T. M., McKeegan, K. D. & Le Fort, P. 1995. Detection of inherited monazite in the Manaslu leucogranite by 208Pb/232Th ion microprobe dating; crystallization age and tectonic implications. Earth and Planetary Science Letters 133, 271–82. Harrison, T. M., Grove, M., Lovera, O. M. & Catlos, E. J. 1998. A model for the origin of Himalayan anatexis and inverted metamorphism. Journal of Geophysical Research 103, 27017–32. Harrison, T. M., Grove, M., McKeegan, K. D., Coath, C. D., Lovera, O. M. & Le Fort, P. 1999. Origin and episodic emplacement of the Manaslu intrusive complex, central Himalaya. Journal of Petrology 40, 3–19. Hauck, M. L., Nelson, K. D., Brown, L. D., Zhao, W. J. & Ross, A. R. 1998. Crustal structure of the Himalayan orogen at similar to 90 degrees east longitude from Project INDEPTH deep reﬂection proﬁles. Tectonics 17, 481–500. Hodges, K. V. 2000. Tectonics of the Himalaya and southern Tibet from two perspectives. Geological Society of America Bulletin 112, 324–50. Hodges, K. V., Parrish, R. R. & Searle, M. P. 1996. Tectonic evolution of the central Annapurna Range, Nepalese Himalayas. Tectonics 15, 1264–91. Hodges, K. V., Bowring, S. A., Davidek, K. L., Hawkins, D. P. & Krol, M. A. 1998. Evidence for rapid displacement on Himalayan normal faults and the importance of tectonic denudation in the evolution of mountain ranges. Geology 26, 483–6. Hubbard, M. S. 1989. Thermobarometric constraints on the thermal history of the Main Central Thrust Zone and Tibetan Slab, eastern Nepal Himalaya. Journal of Metamorphic Geology 7, 19–30. Jackson, J., McKenzie, D., Priestley, K. & Emmerson, B. 2008. New views on the structure and rheology of the lithosphere. Journal of the Geological Society, London 165, 453–65. Jamieson, R. A., Beaumont, C., Medvedev, S. & Nguyen, M. H. 2004. Crustal channel ﬂows: 2. Numerical models with implications for metamorphism in the Himalayan–Tibetan orogen. Journal of Geophysical Research – Solid Earth 109, B06407. Jessup, M. J., Law R. D., Searle, M. P. & Hubbard, M. S. 2006. Structural evolution and vorticity of ﬂow during extrusion and exhumation of the Greater Himalayan Slab, Mount Everest Massif, Tibet/Nepal; implications for orogen-scale ﬂow partitioning. In Law, R. D., Searle, M. P. & Godin, L. (eds) Channel Flow, Ductile Extrusion and Exhumation in Continental Collision Zones. Geological Society, London, Special Publications 268, 379–413. Jessup, M. J., Cottle, J. M., Searle, M. P., Law, R. D., Newell, D. L., Tracy, R. J. & Waters. D. J. 2008. P–T–t–D paths of Everest Series schist, Nepal. Journal of Metamorphic Geology 26, 717–39. doi: 10.1111/j.1525-1314.2008.00784.x Klemperer, S. L. 2006. Crustal ﬂow in Tibet; geophysical evidence for the physical state of Tibetan lithosphere, and inferred patterns of active ﬂow. In Law, R. D., Searle, M. P. & Godin, L. (eds) Channel Flow, Ductile Extrusion and Exhumation in Continental Collision Zones. Geological Society, London, Special Publications 268, 39–70. Law, R. D., Searle, M. P. & Simpson, R. L. 2004. Strain, deformation temperatures and vorticity of ﬂow at the top of the Greater Himalayan Slab, Everest Massif, Tibet. Journal of the Geological Society, London 161, 305–20. Law, R. D., Searle, M. P. & Godin, L. (eds) 2006. Channel flow, ductile extrusion and exhumation in continental collision zones. Geological Society, London, Special Publication 268. 620 pp. Le Breton, N. & Thompson, A. B. 1988. Fluid-absent (dehydration) melting of biotite in metapelites in the early stages of crustal anatexis. Contributions to Mineralogy and Petrology 99, 226–37.

232

M. P. SEARLE ET AL.

LeFort, P. 1975. Himalayas the Collided Range – Present Knowledge of Continental Arc. American Journal of Science A275, 1–44. LeFort, P. 1981. Manaslu Leucogranite – a Collision Signature of the Himalaya; a Model for its Genesis and Emplacement. Journal of Geophysical Research 86, 545–68. Murphy, M. A. & Harrison, T. M. 1999. Relationship between leucogranites and the Qomolangma detachment in the Rongbuk Valley, south Tibet. Geology 27, 831–4. Nelson, K. D., Zhao, W., Brown, L. D., Kuo, J., Che, J., Liu, X., Klemperer, S. L., Makovsky, Y., Meissner, R., Mechie, J., Kind, R., Wenzel, F., Ni, J., Nabelek, J., Chen, L., Tan, H., Wei, W., Jones, A. G., Booker, J., Unsworth, B., Kidd, W. S. F., Hauck, M., Alsdorf, D., Ross, A., Cogan, M., Wu, C., Sandvol, E. A. & Edwards, M. A. 1996. Partially molten middle crust beneath southern Tibet; synthesis of Project INDEPTH results. Science 274, 1684–8. Noble, S. R. & Searle, M. P. 1995. Age of Crustal Melting and Leucogranite Formation from U–Pb Zircon and Monazite Dating in the Western Himalaya, Zanskar, India. Geology 23, 1135–8. Parrish, R. R. 1990. U–Pb dating of monazite and its application to geological problems. Canadian Journal of Earth Sciences 27, 1431–50. Parrish, R. R., Gough, S. J., Searle, M. P. & Waters, D. J. 2006. Plate velocity exhumation of ultrahigh-pressure eclogites in the Pakistan Himalaya. Geology 34, 989–92. Patin˜o Douce, A. E. & Harris, N. B. W. 1998. Experimental constraints on Himalayan anatexis. Journal of Petrology 39, 689–710. Peto¨, P. 1976. An experimental investigation of melting relations involving muscovite and paragonite in the silica-saturated portion of the system K2O–Na2O–Al2O3–SiO2–H2O to 15 kb total pressure. Progress in Experimental Petrology 3, 41–5. Pinet, C. & Jaupart, C. 1987. A thermal model for the distribution in space and time of Himalayan granites. Earth and Planetary Science Letters 84, 87–99. Priestley, K., Jackson, J. & McKenzie, D. 2008. Lithospheric structure and deep earthquakes beneath India, the Himalaya and southern Tibet. Geophysical Journal International 172, 345–62. Prince, C., Harris, N. B. W. & Vance, D. 2001. Fluid-enhanced melting during prograde metamorphism. Journal of the Geological Society, London 158 (2), 233–41. Rai, S. S., Priestley, K., Gaur, V. K., Mitra, S., Singh, M. P. & Searle, M. P. 2006. Conﬁguration of the Indian Moho beneath the NW Himalaya and Ladakh. Geophysical Research Letters 33, L15308. Rowley, D. B. 1996. Age of initiation of collision between India and Asia: A review of stratigraphic data. Earth and Planetary Science Letters 145, 1–13. Scaillet, B., France-Lanord, C. & LeFort, P. 1990. Badrinath– Gangotri Plutons (Garhwal, India) – Petrological and Geochemical Evidence for Fractionation Processes in a High Himalayan Leucogranite. Journal of Volcanology and Geothermal Research 44, 163–88. Scaillet, B., Pecher, A., Rochette, P. & Champenois, M. 1995. The Gangotri Granite (Garhwal Himalaya) – Laccolithic Emplacement in an Extending Collisional Belt. Journal of Geophysical Research – Solid Earth 100, 585–607. Scaillet, B., Holtz, F., Pichavant, M. & Schmidt, M. 1996. Viscosity of Himalayan leucogranites: Implications for mechanisms of granitic magma ascent. Journal of Geophysical Research – Solid Earth 101, 27691–9. Scaillet, B. & Searle, M. P. 2006. Mechanisms and timescales of felsic magma segregation, ascent and emplacement in the Himalaya. In Law, R. D., Searle, M. P. & Godin, L. (eds) Channel Flow, Ductile Extrusion and Exhumation in Continental Collision Zones. Geological Society, London, Special Publications 268, 293–308. Bath, UK: The Geological Society Publishing House. Scha¨rer, U. 1984. The eﬀect of initial 230Th disequilibrium on young U–Pb ages: the Makalu case, Himalaya. Earth and Planetary Science Letters 67, 191–204. Schulte-Pelkum, V., Monsalve, G., Sheehan, A., Pandey, M. R., Sapkota, S., Bilham, R. & Wu, F. 2005. Imaging the Indian subcontinent beneath the Himalaya. Nature 435, 1222–5. Searle, M. P. 1999a. Emplacement of Himalayan leucogranites by magma injection along giant sill complexes: examples from the Cho Oyu, Gyachung Kang and Everest leucogranites (Nepal Himalaya). Journal of Asian Earth Sciences 17, 773–83. Searle, M. P. 1999b. Extensional and compressional faults in the Everest–Lhotse massif, Khumbu Himalaya, Nepal. Journal of the Geological Society, London 156, 227–40. Searle, M. P. 2003. Geological Map of the Mount Everest massif, Nepal – South Tibet Himalaya. Scale 1:100,000. Oxford: Oxford University.

Searle, M. P. 2007. Geological Map of the Mount Everest–Makalu region Nepal – South Tibet Himalaya. Scale 1:100,000. Oxford: Oxford University. Searle, M. P., Cooper, D. J. W. & Rex, A. J. 1988. Collision tectonics of the Ladakh–Zanskar Himalaya. Philosophical Transactions of the Royal Society, London A326, 117–50. Searle, M. P., Pickering, K. T. & Cooper, D. J. W. 1990. Restoration and evolution of the intermontane Indus molasse basin, Ladakh Himalaya, India. Tectonophysics 174, 301–14. Searle, M. P., Crawford, M. B. & Rex, A. J. 1992. Field Relations, Geochemistry, Origin and Emplacement of the Baltoro Granite, Central Karakoram. Transactions of the Royal Society of Edinburgh: Earth Sciences 83, 519–38. Searle, M. P., Metcalfe, R. P., Rex, A. J. & Norry, M. J. 1993. Field relations, petrogenesis and emplacement of the Bhagirathi leucogranite, Garhwal Himalaya. In Treloar, P. J. & Searle, M. P. (eds) Himalayan Tectonics. Geological Society, London, Special Publications 74, 429–44. Bath, UK: The Geological Society Publishing House. Searle, M. P., Parrish, R. R., Hodges, K. V., Hurford, A. J., Ayres, M. W. & Whitehouse, M. J. 1997. Shisha Pangma leucogranite, south Tibetan Himalaya: Field relations, geochemistry, age, origin, and emplacement. Journal of Geology 105, 295–317. Searle, M. P., Waters, D. J., Dransﬁeld, M. W., Stephenson, B. J., Walker, J. D., Walker, C. B. & Rex, D. C. 1999a. Thermal and mechanical models for the structural evolution of the Zanskar High Himalaya. In MacNiocall, C. & Ryan, P. D. (eds) Continental Tectonics. Geological Society, London, Special Publications 164, 139–56. Bath, UK: The Geological Society Publishing House. Searle, M. P., Noble, S. R., Hurford, A. J. & Rex, D. C. 1999b. Age of crustal melting, emplacement and exhumation history of the Shivling leucogranite, Garhwal Himalaya. Geological Magazine 136, 513–25. Searle, M. P., Simpson, R. L., Law, R. D., Parrish, R. R. & Waters, D. J. 2003. The structural geometry, metamorphic and magmatic evolution of the Everest massif, High Himalaya of Nepal – South Tibet. Journal of the Geological Society, London 160, 345–66. Searle, M. P., Law R. D. & Jessup, M. J. 2006. Crustal structure, restoration and evolution of the Greater Himalaya in NepalSouth Tibet; implications for channel ﬂow and ductile extrusion of the middle crust. In Law, R. D., Searle, M. P. & Godin, L. (eds) Channel Flow, Ductile Extrusion and Exhumation in Continental Collision Zones. Geological Society, London, Special Publications 268, 355–78. Bath, UK: The Geological Society Publishing House. Searle, M. P., Law, R. D., Godin, L., Larson, K., Streule, M. J., Cottle, J. M. & Jessup, M. J. 2008. Deﬁning the Himalayan Main Central Thrust in Nepal. Journal of the Geological Society, London 165, 523–34. Searle, M. P. & Godin, L. 2003. The South Tibetan detachment and the Manaslu Leucogranite; a structural reinterpretation and restoration of the Annapurna–Manaslu Himalaya, Nepal. Journal of Geology 111, 505–23. Searle, M. P. & Rex, A. J. 1989. Thermal-Model for the Zanskar Himalaya. Journal of Metamorphic Geology 7, 127–34. Searle, M. P. & Szulc, A. G. 2005. Channel ﬂow and ductile extrusion of the high Himalayan slab – the Kangchenjunga–Darjeeling proﬁle, Sikkim Himalaya. Journal of Asian Earth Sciences 25, 173–85. Simpson, R. L., Parrish, R. R., Searle, M. P. & Waters, D. J. 2000. Two episodes of monazite crystallization during metamorphism and crustal melting in the Everest region of the Nepalese Himalaya. Geology 28, 403–6. Stevens, G., Clemens, J. D. & Droop, G. T. R. 1997. Melt production during granulite facies anatexis: experimental data from ‘primitive’ metasedimentary protoliths. Contributions to Mineralogy and Petrology 128, 352–70. Thimm, K. A., Parrish, R. R., Hollister, L. S., Grujic, D., Klepeis, K. & Dorji, T. 1999. New U–Pb data from the Lesser and Greater Himalayan sequences in Bhutan. Journal of Conference Abstracts 4, 57. Tilmann, F., Ni, J., Hearn, T., Ma, Y. S., Rapine, R., Kind, R., Mechie, J., Saul, J., Haines, S., Klemperer, S., Brown, L., PanAnont, P., Ross, A., Nelson, K. D., Guo, J., Zhao, W. & INDEPTH III Seismic Team. 2003. Seismic imaging of the downwelling Indian lithosphere beneath central Tibet. Science 300, 1424–7. Treloar, P. J., O’Brien, P. J., Parrish, R. R. & Khan, M. a. 2003. Exhumation of Tertiary coesite-bearing eclogites from the

HIMALAYAN CRUSTAL MELT GRANITES AND MIGMATITES Pakistan Himalaya. Journal of the Geological Society, London 160, 367–76. Unsworth, M. J., Jones, A. G., Wei, W., Marquis, G., Gokarn, S. G., Spratt, J. E. & INDEPTH-MT Team. 2005. Crustal rheology of the Himalaya and Southern Tibet inferred from magnetotelluric data. Nature 438, 78–81. Vance, D. & Harris, N. 1999. Timing of prograde metamorphism in the Zanskar Himalaya. Geology 27, 395–8. Vielzeuf, D. & Holloway, J. R. 1988. Experimental determination of the ﬂuid-absent melting relations in the pelitic system; consequences for crustal diﬀerentiation. Contributions to Mineralogy and Petrology 98, 257–76. Viskupic, K., Hodges, K. V. & Bowring, S. A. 2005. Timescales of melt generation and the thermal evolution of the Himalayan metamorphic core, Everest region, eastern Nepal. Contributions to Mineralogy and Petrology 149, 1–21. Viskupic, K. & Hodges, K. V. 2001. Monazite–xenotime thermochronometry: methodology and an example from the Nepalese Himalaya. Contributions to Mineralogy and Petrology 141, 233– 47. Walker, J. D., Martin, M. W., Bowring, S. A., Searle, M. P., Waters, D. J. & Hodges, K. V. 1999. Metamorphism, melting, and extension: Age constraints from the High Himalayan Slab of southeast Zanskar and northwest Lahaul. Journal of Geology 107, 473–95. Walker, C. B., Searle, M. P. & Waters, D. J. 2001. An integrated tectonothermal model for the evolution of the High Himalaya in western Zanskar with constraints from thermobarometry and metamorphic modeling. Tectonics 20, 810–33. Wei, W., Unsworth, M., Jones, A., Booker, J., Tan, H., Nelson, D., Chen, L., Li, S., Solon, K., Bedrosian, P., Jin, S., Deng, M., Ledo, J., Kay, D. & Roberts, B. 2001. Detection of widespread ﬂuids in the Tibetan crust by magnetotelluric studies. Science 292, 716–18. Weinberg, R. F. & Searle, M. P. 1999. Volatile-assisted intrusion and autometasomatism of leucogranites in the Khumbu Himalaya, Nepal. Journal of Geology 107, 27–48. White, R. W., Powell, R. & Holland, T. J. B. 2001. Calculation of partial melting equilibria in the system Na2O–CaO–K2O–FeO– MgO–Al2O3–SiO2–H2O (NCKFMASH). Journal of Metamorphic Geology 19 (2), 139–53.

233

Whittington, A. G., Harris, N. B. W. & Baker, J. 1998. Low-pressure crustal anatexis: the signiﬁcance of spinel and cordierite from metapelitic assemblages at Nanga Parbat, northern Pakistan. In Treloar, P. J. & O’Brien, P. J. (eds) What Drives Metamorphism and Metamorphic Reactions? Geological Society, London, Special Publications 138, 183–98. Bath, UK: The Geological Society Publishing House. Whittington, A. G., Harris, N. B. W. & Butler, R. W. H. 1999. Contrasting anatectic styles at Nanga Parbat, northern Pakistan. In Macfarlane, A., Sorkhabi, R. B. & Quade, J. (eds) Himalaya and Tibet: Mountain Roots to Mountain Tops. Geological Society of America, Special Paper 328, 129–44. Whittington, A. G. & Treloar, P. J. 2002. Crustal anatexis and its relation to the exhumation of collisional orogenic belts, with particular reference to the Himalaya. Mineralogical Magazine 66, 53–91. Wu, C., Nelson, K. D, Wortman, G., Samson, S., Yue, Y., Li, J., Kidd, W. S. F. & Edwards, M. A. 1998. Yadong cross-structure and South Tibetan Detachment in the east central Himalaya. Tectonics 17, 28–45. Zeitler, P. K., Koons, P. O., Bishop, M. P., Chamberlain, C. P., Craw, D., Edwards, M. A., Hamidullah, S., Jan, M. Q., Khan, M. A., Khattak, M. U. K., Kidd, W. S. F., Mackie, R. L., Meltzer, A. S., Park, S. K., Pecher, A., Poage, M. A., Sarker, G., Schneider, D. A., Seeber, L. & Shroder, J. F. 2001a. Crustal reworking at Nanga Parbat, Pakistan: Metamorphic consequences of thermalmechanical coupling facilitated by erosion. Tectonics 20, 712–28. Zeitler, P. K., Meltzer, A. S., Koons, P. O., Craw, D., Hallet, B., Chamberlain, C. P., Kidd, W. S. F., Park, S. K., Seeber, L., Bishop, M. & Shroder, J. 2001b. Erosion, Himalayan geodynamics and the Geomorphology of metamorphism. GSA Today 11 (January) 4–9. Zhang, H. F., Harris, N. B. W., Parrish, R. R., Kelley, S., Zhang, L., Rogers, N., Argles, T. & King, J. 2004. Causes and consequences of protracted melting of the mid-crust exposed in the North Himalayan antiform. Earth and Planetary Science Letters 228, 195–212. Zhu, B., Kidd, W. S. F., Rowley, D. B., Currie, B. S. & Shaﬁque, N. 2005. Age of initiation of the India–Asia collision in the eastcentral Himalaya. Journal of Geology 113, 265–85.

MS received 11 December 2007. Accepted for publication 13 March 2008 (Stellenbosch); 15 January 2009 (RSE).

Earth and Environmental Science Transactions of the Royal Society of Edinburgh, 100, 235–249, 2010 (for 2009)

Zoning and resorption of plagioclase in a layered gabbro, as a petrographic indicator of magmatic diﬀerentiation Takashi Hoshide and Masaaki Obata Division of Earth and Planetary Sciences, Department of Geology and Mineralogy, Graduate School of Science, Kyoto University, Kyoto, Japan 606–8502 Email: [email protected]; [email protected] ABSTRACT: The Murotomisaki Gabbroic Intrusion is a sill-like layered gabbro emplaced in sedimentary strata of Tertiary age in southwest Japan. The zoning (including resorption structures) and the compositional variations of plagioclase from throughout the intrusion were studied, and it was found that the zoning pattern may be classiﬁed into four types, which may well correlated with the hosting rock types, the mode of occurrences and their stratigraphic positions in the intrusion. The plagioclase zoning was successfully decoded, and the sequence of events that took place during the magmatic diﬀerentiation was deduced and further interpreted in the context of a stratiﬁed basal boundary layer slowly ascending in a solidifying magma body. It was shown that various layered structures – modal layering, podiform gabbroic pegmatites and anorthositic layers – observed in the Murotomisaki Gabbro were formed within the moving basal boundary layer by ﬂushing of H2O-rich ﬂuid and fractionated silicate melts from below. By the ﬂuxing of hydrous ﬂuids, plagioclase crystals preferentially dissolved and then melt fraction increased in the basal boundary layer. Under these circumstances, plagioclase-rich fractionated melts diapirically segregated from the crystal pile. Calcic plagioclases, which are out of equilibrium in the central part of the intrusion, may have originated from the basal boundary layer in this manner. KEY WORDS:

Anorthosite, boundary layer, crystal mush, diapir, ﬂuid ﬂuxing, magma chamber

Magma reservoirs are subjected to crystallisation and associated fractionation that are driven by the various physicochemical processes such as gravitational settling or ﬂotation of crystals, compositional convection and compaction. As thermal gradient is created near the marginal zones of the magma reservoir (e.g. Brandeis & Jaupart 1986; Turner et al. 1986; Marsh 1988; Tait & Jaupart 1992), it is plausible that processes in the magmatic boundary layer play an important role in the magmatic diﬀerentiation after the settling of phenocrysts originally contained in the emplaced magma (Mangan & Marsh 1992). Recent studies of crystal size and crystal number density of olivine by Hoshide et al. (2006a, b) identiﬁed the zone of crystal accumulation (AC subzone) and the zone of crystal growth (GR subzone). On the basis of the vertical variation and the numerical balance of crystal number density, they thought that crystal settling hardly occurred and some sort of processes involved in in situ crystallisation (crystal growth) were important in the magmatic diﬀerentiation after the major phase of crystal settling of olivine originally contained in the magma at the time of its emplacement. Furthermore, they emphasised the occurrence of anorthosites in the GR subzone and proposed a hypothesis that segregation and diapiric ascent of anorthositic crystal-mush played an important role in the formation of the GR subzone and thereby aﬀected the entire magmatic evolution. The argument is largely based on the observed whole-rock compositional trends and plume-like occurrence of the anorthosites. The present paper takes a diﬀerent approach to studying the magmatic processes in more detail, by looking at rock microstructure, particularly, the zonal structure of plagioclase from the Murotomisaki Gabbro. In contrast to ferromagnesian minerals like olivine or augite,

2009 The Royal Society of Edinburgh. doi:10.1017/S1755691009016090

plagioclase has the advantage of slow diﬀusion kinetics, so that original zoning formed during crystal growth, i.e., growth zoning, is well preserved in the crystals, thereby providing useful information on the sequence of events and temporal change of physico-chemical environment in the magmatic system. Following the geological outline and petrographic description, analytical data on plagioclase zoning are presented, and it is shown that the zoning pattern is well correlated to the hosting rock types and the mode of occurrence. The zoning data and rock systematics will be interpreted in the context of a moving and crystallising boundary layer. It is emphasised that the zoned plagioclase may be used as a useful tracer of mass movement within a magmatic system under diﬀerentiation.

1. Geological setting The Murotomisaki Gabbroic Intrusion (abbr. ‘MGI’) is a wedge-shaped, sill-like layered gabbro (up to 220 m thickness) located near Cape Muroto, Shikoku, Japan. This intrusion is emplaced nearly concordantly to bedded sandstone and mudstone of the Tertiary sediments (Shimanto Group), giving rise to contact metamorphism on both sides of the sill. The intrusion has been dated at 14 Ma by the Rb–Sr whole-rock isochron method applied to anatexites in the contact aureole produced by the hot magma emplacement (Hamamoto & Sakai 1987). Regional geological and palaeomagnetic studies show that the intrusion was nearly horizontal when the magma was emplaced and solidiﬁed (Yajima 1972; Kodama et al. 1983; Kodama & Koyano 2003). The intrusion, together with

236

TAKASHI HOSHIDE AND MASAAKI OBATA

Figure 1 Lithological and compositional variations across the Murotomisaki Gabbroic Intrusion: (a) lithological variation; (b) modal variation; (c) crystal size and crystal number density of olivine; (d) whole-rock MgO content. Modiﬁed from Hoshide et al. (2006a). The crystal number density of olivine is shown in logarithmic scale. A broken line in (c) shows the level of average value of crystal number density of olivine of the intrusion. A broken line in (d) is the average value of the whole-rock MgO. Abbreviations: LZ=Lower Zone; MZ=Middle Zone; UZ=Upper Zone; PGP=podiform gabbroic pegmatite; AnL=anorthositic layer. Pl=plagioclase; Ol= olivine; Aug=augite; Opq=opaque minerals; Oth=other phases (hypersthene, biotite, hornblende and chlorite).

the surrounding sedimentary strata, was tilted by later tectonics by 60–70( towards northwest. A near-complete crosssection of the intrusion is exposed along the sea coast (Fig. 1a).

2. Petrographic subdivisions of the MGI According to the rock types and the mesoscopic layered structures observable in the ﬁeld, the intrusion is divided into the following three major stratigraphic zones: the Lower Zone (LZ: about 105 m thick); the Middle Zone (MZ: about 75 m thick); and the Upper Zone (UZ: about 30 m thick) (Fig. 1a). The Lower Zone and the Upper Zone contain thin chilled margins of a few metres thick. The Lower Zone consists of ﬁne- to medium-grained olivine gabbro and contains, in places, subordinate amounts of anorthositic layers (Fig. 2a) and podiform gabbroic pegmatites (Fig. 3). The Middle Zone mainly consists of coarse gabbro that is composed of augite and plagioclase, and characteristically lacks olivine. The Upper Zone consists of medium-grained olivine gabbro, whose textures are diﬀerent from those of Lower-Zone olivine gabbros. Important petrographic features are summarised in Figure 1 and described below.

2.1. The Lower Zone Using the olivine crystal number density (i.e., total number of crystals of olivine per unit volume of rocks), the Lower Zone

has been subdivided into two subzones, the AC subzone and the GR subzone above (Fig. 1). The AC subzone comprises the lower 40m and has crystal number density of olivine above the average of the whole intrusion, whereas the GR subzone (40–100 m) is less than the average. The crystal size of olivine and the presence or absence of the layered structures are clearly diﬀerent between the AC and GR subzones. Therefore, the ﬁeld occurrence and petrography are described separately for each subzone. 2.1.1. The AC subzone. The AC subzone appears uniform in the ﬁeld; the grain size increases gradually upwards from the lower contact. The chilled margin, within a few cm from the contact, is very ﬁne-grained, containing a few microphenocrysts of olivine (up to 8 vol%, average grain size 0·25 mm) and plagioclase (w1–3 vol%, average grain size 0·1 mm) in a darkcoloured matrix (probably a devitriﬁed glass) that contains small skeletal or swallow-tailed crystals of plagioclase (average grain size 10 m) (Fig. 4a). Olivine increases rapidly in both mode and number density with distance from the lower contact, and the maximum is observed at ca. 15 metres (peak O1) before decreasing (Fig. 1b, c). Crystal size of olivine, on the other hand, remains nearly constant. Plagioclase monotonically increases in size, which results in an overall upward-coarsening of rocks. The texture is typically poikilitic. Euhedral to subhedral plagioclase partially encloses euhedral olivine. Anhedral augite

ZONING OF PLAGIOCLASE IN A LAYERED INTRUSION

237

Figure 2 Field photographs of the Murotomisaki Gabbro: (a) Anorthositic layers (AnL, light coloured) in the olivine–gabbro matrix of the GR subzone (approximately 78 m from the bottom). Note a ﬂat bottom and a wavy upper surface. Arrow indicates the top of the sill; (b) closeup of (a). Thin maﬁc selvage attached to the base of an anorthositic layer. The layer trails down in a wavy form; (c) Modally graded layering in the lowermost part of the GR subzone (approximately 40 m from the bottom); (d) Field photograph of a coarse gabbro in the Middle Zone (approximately 150 m from the bottom) showing a mesoscopic patchy structure, consisting of melanocratic patches rich in augite and opaque mineral and leucocratic plagioclase-rich matrix.

encloses olivine and plagioclase or occupies their interstices (Fig. 4b). Plagioclase usually has a euhedral calcic core nearly constant in An-content and this An-content decreases continuously towards the rim (Fig. 5a). The plagioclase cores often show oscillatory zoning and have thin edges of even more calcic composition than the core. More details of the zonal structures are given in the next section. The olivine crystals often contain small polycrystalline clots (several tens m in size) consisting of pargasitic amphibole, hypersthene and biotite that are hereafter called ‘amphibole clots’ (Fig. 5d). Many clots are nearly-round, but some have the form of a ‘negative crystal’ of the host olivine. The morphology and the occurrence suggest that these clots probably represent solidiﬁed hydrous silicate melt entrapped in growing olivine crystals at magmatic stages. 2.1.2. The GR subzone. The GR subzone is lithologically more variable. The observed modal range is 12–40% olivine, 5–30% augite and 52–65% plagioclase; olivine grain size varies from 1·0 mm to 2·5 mm. Compositional layering is more developed than in the AC subzone (Fig. 1). This zone contains abundant layers of anorthosite (Fig. 2a) and lenticular pods of gabbroic pegmatite (Fig. 3a, b). There are two modal peaks of olivine in the host gabbro (O2 and O3), where average grain size of olivine is maximum. These two peaks coincide in

position with the minimum peaks of olivine number density. In addition, peak O2 coincides with the maximum whole-rock MgO content (Fig. 1). In olivine gabbros of the GR subzone, euhedral to subhedral plagioclase partially encloses euhedral olivine in common with that of the AC subzone, but olivines are larger and have lower number density than in the AC subzone and thus the texture is not poikilitic. Around the peak O2, augite is less abundant (below 10 vol%) and ﬁlls interstices between cumulus olivine and plagioclase (Fig. 4c). Gabbros around the peak O2 have modal ratios of olivine/plagioclase universally higher than those around peak O3. Around peak O3, olivine crystals are large and irregularly shaped and typically enclose plagioclase. The amphibole clots in the olivine crystals (Fig. 5d) are more abundant and are larger in size (w1 mm) than those of the AC subzone. The amphibole clots also occur at interstices of aggregates of olivine crystals. A ﬁne-scale rhythmic layering due to modal variation of plagioclase, olivine and augite (with typical wave length around 10 cm) locally develops in a lowermost part of the GR subzone (the ‘compositional banding’ of Yajima 1972) (Fig. 2c). The felsic layers show weak shape-preferred orientations of plagioclase parallel to the layering. The mode of the layering is vertically asymmetric, that is, a felsic layer gradually changes

238

TAKASHI HOSHIDE AND MASAAKI OBATA

into the overlying maﬁc layer, but sharply into the underlying one. The modally-layered portion is more coarse-grained than the surrounding massive and more homogeneous parts at the same level. Plagioclase varies in the modal amount by 15% to 60% in this layering. Plagioclase and olivine are cumulus phases, and augite occurs at interstitial positions. Plagioclase crystals have calcic cores, the shapes of which are slightly rounded or corroded compared to those in non-layered gabbros. In the felsic layers, their calcic cores are larger than those in the maﬁc layers. 2.1.3. Bimodal occurrence of anorthosites (in the ‘GR subzone’). Two types of anorthosite occur in the GR subzone. One type forms a constituent part of podiform gabbroic pegmatites (Fig. 3a, b); the other forms individual mediumgrained layers (Plagioclase grain size: 2 mm in length; w50 cm thick) without a pegmatitic part (Fig. 2a). The anorthositic layer is abundant at the 70–80–m horizon, around the O2 peak, whereas the podiform gabbroic pegmatite is more abundant just below and above this horizon (Fig. 1a). Podiform gabbroic pegmatite. The podiform gabbroic pegmatites are lenticular in shape, oriented sub-parallel to the layering and vary in thickness from a few cm to 2 m (Fig. 3). They form clusters of several metres to several tens of metres in size (Fig. 3c). Each pod typically has an internal layered structure consisting of an anorthositic roof, a pegmatitic ﬁlling, and a picritic ﬂoor (Fig. 3b). The anorthositic roof is medium- to coarse-grained (plagioclase grain size, about 3–10 mm in length) and has strong shape fabrics of plagioclase; euhedral plagioclase crystals are aligned sub-horizontal, parallel to the lower and the upper boundaries. The upper boundary of the anorthositic roof against the host olivine gabbro is often wavy and the plagioclase crystals are aligned parallel to the wavy boundary. This wavy structure has locally developed into plume-like, ‘diapiric structure’, implying a gravitational ﬂuid-dynamic instability being developed at magmatic stages (Fig. 3d). Globular small masses of anorthosite (about 10 cm in diameter) occur above these plumes (Fig. 3d, f). The anorthositic roof and the ‘globular’ anorthosite masses contain abundant clots of amphibole of a few mm in diameter (Fig. 3e). Near the margin of the globular anorthosite masses plagioclase crystals are also aligned roughly parallel to their boundaries. Microscopically, the anorthosite in the roof has an ophitic texture, consisting of euhedral plagioclase crystals (60– 80 vol%) enclosed in large augite crystals (20–35 vol%) (Fig. 4g). Some plagioclase crystals show undulose extinction, implying plastic deformation of crystals. They also contain small amounts of calcite, apatite, titanite and allanite that occur in the interstices of euhedral plagioclase. Other accessory phases are ilmenite, titaniferous magnetite and sulﬁdes. Intergrowths of ilmenite with titanite or titaniferous magnetite are common in the anorthositic roofs, whereas these oxides occur as independent grains in the host olivine gabbro. The pegmatitic ﬁlling is very coarse-grained and consists of euhedral plagioclase (about 1–4 cm in length; wAn30–50) and interstitial augite. Plagioclase crystals are aligned nearly perpendicular to the sharp boundary with the anorthositic roof (Fig. 3a). The picritic ﬂoor is also medium-grained (olivine grain size: about 2–3 mm in length) and consists of euhedral olivine (40w70 vol%) and interstitial plagioclase and augite (Fig. 4h). Olivine in the picritic gabbro in the ﬂoor is identical in composition to that in the host olivine gabbro (Mg#=60–65). An outstanding feature is the invariability of the sequence of the internal layering with the anorthositic roof always on the top and the picritic ﬂoor on the bottom. The volume proportion of these three parts varies; the anorthositic roof may

be very thin in some podiform pegmatites (Fig. 3b), whereas the picritic ﬂoor is minor or even missing in others (Fig. 3a). Anorthositic layers. Anorthositic layers vary in thickness between 10 cm and 1 m and are oriented parallel to the compositional layering. The upper boundary to the host gabbro is wavy, like that of the anorthositic roof of the podiform gabbroic pegmatite, and the lower boundary is ﬂat (Fig. 2a). Anorthosite has strong crystal shape fabrics, with plagioclase crystals aligned parallel to the wavy contacts (see also ﬁg. 3B in Hoshide et al. 2006b). Some anorthositic layers have thin discontinuous basal layers of medium-grained maﬁc rocks (olivine grain size: about 2–3 mm in length), like the picritic ﬂoor of the podiform gabbroic pegmatite, implying its genetic link with podiform gabbroic pegmatites (Fig. 2b). The texture and rock composition of the anorthositic layers are very similar to the roof anorthosite.

2.2. The Middle Zone The Middle Zone lacks obvious layering, but characteristically contains patchy maﬁc domains of a few to tens of centimetres in size (Fig. 2d). Euhedral plagioclase crystals included in large augite grains (up to several centimetres in size) form an ophitic texture (Fig. 4d, e). Olivine is generally absent except in one sample (at the 152-m level). The olivine in this sample is very iron-rich and shows a reverse zoning from an Fe-rich core (wFo37) to an Fe-poor rim (wFo43). Hydrous minerals, such as hornblende, biotite and prehnite, occur as interstitial phases, as well as in the ‘embayments’ of the calcic cores of plagioclase. Some rocks from the lower part of the Middle Zone have large plagioclase crystals that ﬁll interstices of euhedral to subhedral augite crystals. A large block (15 m across) of medium-grained olivine gabbro occurs within the Middle Zone. The boundary between the block and the surrounding coarse gabbro is not clear because of poor exposure.

2.3. The Upper Zone Olivine gabbro, which consists of olivine, plagioclase and augite, is the main rock type in this zone. Starting from the upper contact, olivine rapidly increases downward in the modal abundance and in crystal number density, marking a sharp peak (O4) at about the 210-m level (2 m below the upper contact; Fig. 1). The olivine grain size near the contact is about the same as that of AC subzone in the Lower Zone, but increases rapidly below the modal peak O4 and reaches a maximum at about the 206-m level (Fig. 1c). Around peak O4 there is a horizon that contains abundant spherulitic clots of hornblende aggregate (about 0·2 mm to 3 cm in diameter). Although the host rocks are mineralogically similar to the olivine gabbros in the Lower Zone, the texture is diﬀerent. Whilst the plagioclase is euhedral to subhedral, olivine and augite are irregular in shape and ﬁll the interstices of plagioclase crystals in the Upper Zone. Some plagioclase crystals contain subhedral to anhedral calcic cores nearly constant in An-content and the boundary between the calcic core and the sodic mantle is rather sharp, marking a compositional gap as described in next section. Other grains in the same sample do not contain such distinct calcic cores, but show more gradual normal zoning. Plagioclase crystals become aggregated and form cancellous plagioclase-rich networks at the centimetrescale and are oriented in the same direction in the networks. In the networks or clusters (a few centimetres scale), plagioclase crystals show undulose extinction and deformation twins, implying that they have been plastically deformed (Fig. 4f).

ZONING OF PLAGIOCLASE IN A LAYERED INTRUSION

Figure 3 Photographs of podiform gabbroic pegmatites in the GR subzone. Arrow in each photograph indicates the top of the sill: (a) and (b) podiform gabbroic pegmatite (at approximately 40 m from the bottom); (c) Swarming podiform gabbroic pegmatites (at approximately 40 m from the bottom); (d) plume-like structure of anorthosite developed on the roof of podiform gabbroic pegmatite (approximately 50 m from the bottom); (e) An amphibole clot (arrow) in an anorthositic roof. Note plagioclase crystals aligned parallel to the outline of the amphibole clot; (f) A globular mass of anorthosite lying just above the anorthositic roof of the podiform gabbroic pegmatite. The anorthositic mass contains abundant hornblendes at the top of the mass (pointed by a ﬁnger).

239

240

TAKASHI HOSHIDE AND MASAAKI OBATA

Figure 4 Photomicrographs of the Murotomisaki gabbroic rocks: (a) the lower chilled margin, sample LCM32 (just contact with the country rocks); (b) Lower-Zone picritic gabbro, sample 91080503 (AC subzone, 15 m from the bottom, around O1 in Fig. 1); (c) Lower-Zone olivine gabbro, sample 91080706 (around O2 in Fig. 1, at 71 m); (d) Middle-Zone coarse gabbro, sample 91080906 (152 m from the bottom); (e) Middle-Zone coarse gabbro, sample 91081003 (162 m from the bottom); (f) Upper-Zone, sample 92081007 (203 m from the bottom). Plastically-deformed plagioclase crystals showing undulose extinction; (g) anorthositic roof of the podiform gabbroic pegmatite, sample 06050402 (53 m from the bottom); (h) picritic ﬂoor of the podiform gabbroic pegmatite, sample 06080603 (60 m from the bottom). Abbreviations for minerals are the same as in Figure 1.

ZONING OF PLAGIOCLASE IN A LAYERED INTRUSION

241

Figure 5 (a) Oscillatory zoned plagioclase of the Lower-Zone picritic gabbro (sample 91080503). (b) A high-An edge (bright part) occurring around the calcic core of plagioclase in the GR subzone. Note that boundary between the core and the edge is rough. (c) Embayments in the calcic cores that are ﬁlled with sodic plagioclase associated with small amount of hornblende, augite and chlorite. Thin small high-An edges are developed surrounding the embayment. (d) ‘Amphibole clots’ included in an olivine crystal of a Lower-Zone olivine gabbro. Abbreviations: Bt=biotite; Par=pargasite; Spl=spinel. Abbreviations for the other minerals are the same as in Figure 1.

3. Plagioclase zoning and its stratigraphic variations 3.1. Plagioclase zoning The plagioclase-zoning pattern was studied in details using an EDS microprobe at Kyoto University (Hitachi S3500H scanning electron microscope equipped with an EDAX energy dispersive analytical system) and a WDS microprobe at Kobe University (JEOL JXA8900). The accelerating voltage and beam current were maintained at 20 kV and 500 pA at Kyoto, and at 15 kV and 12 nA at Kobe. It was conﬁrmed that the calcic core of plagioclases from the Lower Zone is nearly constant in An-content throughout the core (see below). Some plagioclase cores from the lower part of the Lower Zone (i.e. the AC subzone) show oscillatory zoning (Fig. 5a), but the oscillation tends to fade away in the upper parts. Olivine crystals are typically in contact with the outlines of the calcic cores of plagioclases (Figs 4c, 5a, 6a). The calcic core is typically surrounded by a narrow edge (of 20–100 m width) of even more calcic composition (wAn72–80). This narrow edge of high An-content is referred to as ‘high-An edge’ hereafter (Figs 4c, 5a–c, 6a). The outlines of the calcic cores (i.e. the boundaries between the core and the high-An edge) are nearly straight and smooth (Fig. 4c), but some of

them are rough or wavy (Fig. 5b). Across the core to the high-An edge, the An-content abruptly increases and then rapidly decreases back to about the same An-content level as the core with a steep gradient (Fig. 7a). Furthermore, the gradient discontinuously changes and then the An-content continuously decreases outwards with a gentler gradient. This outer marginal zone of the gentle compositional gradient is referred to as ‘mantle’ hereafter. This zoning pattern, which is typically observed in the host olivine gabbros, is designated as Type A (Fig. 8). Locally, there are embayments or ‘pools’ in the calcic cores of the Type A plagioclases that are ﬁlled with distinctly sodic plagioclase (wAn50) associated with small amount of augite and hornblende (Figs 5c, 6a). In some cases, narrow high-An edges surrounding the calcic cores are developed surrounding the embayment (Fig. 5c). Small grains of hydrous phases such as amphibole and chlorite occur in the embayments of the calcic cores (Fig. 5c) or along the boundaries between the calcic cores and the high-An edges (Fig. 5b). Such embayments or resorption structures are more pervasive and better developed in the anorthosites as described below. The plagioclases in the anorthosites, both from the roofs of the podiform gabbroic pegmatites and the anorthositic layers, have similar zonal structure, but this diﬀers from that of the host gabbros. They have the calcic cores nearly constant in

242

TAKASHI HOSHIDE AND MASAAKI OBATA

Figure 6 Back scatter electron images of plagioclase crystals from various rock types of the Murotomisaki Gabbro. Dashed lines in the BSE images indicate the zoning proﬁle sections in Figure 7: (a) Plagioclase from the Lower Zone (GR subzone, 71 m from the bottom, sample 92080706); (b) Plagioclase from the anorthositic roof of the podiform gabbroic pegmatite (45 m from the bottom, sample 06080703). The narrow high-An edge (arrow) is partly retained where the calcic core is not embayed; (c) Plagioclase from the picritic ﬂoor of the podiform gabbroic pegmatite; (d) Plagioclase from the modal layering (about 40 m from the bottom, sample ML1); (e) Plagioclase from the coarse gabbro in the Middle Zone (152 m from the bottom, sample 91080906); (f) Plagioclase from coarse gabbro in the Middle Zone (162 m from the bottom, sample 91081003). Abbreviations: Hb=hornblende; Chl=chlorite. Abbreviations of the other minerals are the same as in Figure 1.

An-content, which are partly resorbed (Fig. 6b). The resorption structure is more conspicuous in the anorthosites than in the host gabbros. The high-An edges partly surround the outlines of the calcic cores. The An-content abruptly decreases from the core (or the high-An edge) to the mantle, marking a

distinct compositional gap (typically by 15–20 An%) between the core and the mantle (Fig. 7b). Throughout the mantle, the An-content gently decreases outwards. This type of zoning is designated as ‘Type B’ (Fig. 8). The sodic plagioclase that ﬁlls the embayments of the calcic core often includes small grains

ZONING OF PLAGIOCLASE IN A LAYERED INTRUSION

243

Plagioclases from the modal layering (in the lowermost part of the GR subzone) also have the resorbed calcic cores, like those from the anorthosites of the podiform gabbroic pegmatites and the anorthositic layers. The cores are partly surrounded by the high-An edge and have a compositional gap with a distinctly sodic mantle (therefore Type B) (Fig. 6d). However, the calcic cores are less resorbed and the compositional gap between the core and the mantle is smaller (10– 15 An%) than those of the adjacent anorthosites (Figs 6d, 7d). The plagioclase from the modally layered gabbros is, therefore, regarded to be intermediate in terms of zoning and composition between those from the host gabbros and from the anorthosites. Middle-Zone and Upper-Zone gabbros contain at least two types of plagioclase crystals with distinct zoning patterns. One type has irregular-shaped calcic core with nearly constant An-content surrounded by distinctly sodic mantles (Figs 6e, 7e). Some of them can be classiﬁed into Type B, but others lacks the high-An edges. Therefore, the latter zoning pattern is termed as Type B# to distinguish it from Type B (Fig. 8). It is possible that the high-An edges have been lost by extensive resorption, in which case Type B# is simply a derivative from Type B. The other type plagioclase is more sodic in composition and does not contain calcic cores with constant Ancontent (Figs 6f, 7f). It shows normal and gradual zoning and, therefore, is classiﬁed as Type C (Fig. 8). In summary there is a good correlation between the zoning type and host the rock type as illustrated in Figure 8 and Table 1. The Lower-Zone olivine gabbro is dominated by Type A, the anorthosites from the Lower Zone are characterised by Type B zoning, whilst both the picritic ﬂoor of the podiform gabbroic pegmatite and the maﬁc part of the modal layering show Type C. The Middle and Upper Zones contain two types, B and C mixed together, and never contain Type-A plagioclase.

3.2. Stratigraphic variation of plagioclase composition

Figure 7 Zoning proﬁles of plagioclase crystals from various rock types of the Murotomisaki Gabbro: (a) Plagioclase from the Lower Zone (GR subzone, 71 m from the bottom, sample 92080706). Across the core to the high-An edge, (1) the An-content abruptly increases and then (2) rapidly decreases back to about the same An-content as the core with a steep gradient and then (3) continuously decreases outwards with gentler compositional gradients; (b) Plagioclase from the anorthositic roof of the podiform gabbroic pegmatite (45 m from the bottom, sample 06080703). The ‘innermost mantle’ shows the right outside of the calcic core; (c) Plagioclase from the picritic ﬂoor of the podiform gabbroic pegmatite; (d) Plagioclase from the modal layering (about 40 m from the bottom, sample ML1); (e) Plagioclase from the coarse gabbro in the Middle Zone (152 m from the bottom, sample 91080906); (f) Plagioclase from coarse gabbro in the Middle Zone (162 m from the bottom, sample 91081003).

of ferrohornblende, ferro-augite, Fe-chlorite, ilmenite and titaniferous magnetite (Fig. 6b). Plagioclase in the picritic ﬂoors of the podiform gabbroic pegmatites is more sodic (An43–55) than that in the host olivine gabbros and the anorthosites (Fig. 7c). They show nearly constant An-content throughout the crystal but locally have narrow sodic rims without calcic cores like those of Types A or B (Fig. 6c). This zoning type is designated as ‘Type C’ (Fig. 8). The An-content of the core is comparable to that of the innermost mantle (i.e., next to the high-An edge) of the Type B plagioclase from the anorthositic roof within the same podiform gabbroic pegmatites (Figs 7b, 9).

Figure 9 shows a summary of the stratigraphic variation of the An-content of the plagioclase obtained by the microprobe analysis from the host gabbros and the anorthosites (both the anorthositic roofs of the podiform gabbroic pegmatites and the anorthositic layers). In the AC subzone, the An-content of the core from the host gabbro is nearly constant (in the range of An 70–73), whereas in the upper half of the Lower Zone (i.e., the GR subzone), it gradually decreases with increasing stratigraphic height, from An 73 to about An 62. Note that the plagioclase core composition from the anorthosites is nearly the same as those from host olivine gabbro, except one sample at the 78-m level (No. 91040602). This exceptional anorthositic layer has about 5 An mole % higher cores than those from its host gabbro. The An contents of the high-An edges of the plagioclase core, which are constantly about 10 An % higher than the core values, also decrease with increasing stratigraphic height, again with the exception of the anorthosite layer mentioned above. The An content of the innermost parts of the plagioclase mantles (marked ‘im’ in Fig. 9), just outside the high-An edge, (Type B and B# plagioclase) is about 15–20 mole % lower than those of the cores and it also roughly decreases upward. In the Middle-Zone coarse gabbros, there is no clear correlation between the core composition and the stratigraphic height, although the data are sparse in this zone. Out of six samples analysed, four samples contain only plagioclase crystals of Type C zoning and their cores are all less calcic (An 50–69) than the Lower-Zone core compositions. The other two samples (at 152 and 153 m from the bottom) contain plagioclase of Type B or B# zoning. These samples are very heterogeneous with respect to core composition, showing wide

244

TAKASHI HOSHIDE AND MASAAKI OBATA

Figure 8 Schematic illustrations of plagioclase compositional zoning patterns observed from the Murotomisaki Gabbro. Each picture of zoned plagioclase is accompanied by a schematic drawing of the line proﬁle. Table 1 Four types of plagioclase compositional zoning patterns correlated with diﬀerent rock types of the Murotomisaki Gabbro. LZ Modal layering Host gabbro Type A

Type B

O

Podiform gabbro pegmatite

felsic part

mafic part

roof

O

floor

Anorthositic layer

MZ

O

Type B# Type C

pegmatite

UZ

=common O=present but rare.

compositional ranges of An 63–70 and An 53–68, respectively (Fig. 9). The high-An edges of these samples are about An 74, which is about An 8–10% higher than the core value. In the Upper-Zone olivine gabbros, there is also no clear correlation between the core composition and the stratigraphic height. The maximum An value of the core composition (of Type B), however, seems to increase with stratigraphic height toward the upper chilled margin. Two samples from the middle of the Upper Zone (at 199 and 203 m levels) are particularly heterogeneous, containing plagioclases of Type B and Type C (Fig. 8).

4. Discussion on the origin of plagioclase zoning 4.1. Lower Zone The sequence of events inferred from the zonal structure of plagioclase (Type A and Type B in particular) from the Lower

Zone, is (1) the formation of homogeneous calcic plagioclase (though some exceptionally show oscillatory zoning); (2) partial resorption of the calcic plagioclase accompanied (or followed) by the formation of the high-An edges; and (3) the formation of sodic mantle by plagioclase overgrowth. This scheme applies to all the plagioclase throughout the lower zone, both in the host olivine gabbros and the anorthosites, but the extent of resorption is greater in the latter. The latter is also characterised by a distinct compositional gap between the calcic core and the mantle (i.e., Type B, or B#), which may bear important implications for the origin of these structures. The processes of these events will be discussed sequentially below. 4.1.1. Formation of calcic plagioclase and the origin of the stratigraphic variations. The compositional constancy of the plagioclase core observed in the AC-subzone gabbros, including the ‘phenocryst’ in the lower chilled margin (Fig. 9), indicates that the calcic-core parts represent the original

ZONING OF PLAGIOCLASE IN A LAYERED INTRUSION

245

Figure 9 Stratigraphic variations of the plagioclase compositions throughout the Murotomisaki Gabbroic Intrusion. im=innermost mantle. The data of AC subzone are from Akatsuka et al. (1999). An exceptional sample, No. 91040602, is indicated by an arrow (see text).

plagioclase that was already present in the initial magma. It is remembered here that the size of the calcic core from the AC subzone is nearly the same as that of the phenocrysts in the chilled margin. The AC subzone was originally deﬁned as the zone of crystal accumulation of olivine (Hoshide et al. 2006a), but the accumulation hypothesis may also apply to the plagioclase. The sodic mantle surrounding the calcic core, on the other hand, is considered to represent a postcumulus overgrowth as discussed below. The vertical stratigraphic gradient of the plagioclase core composition in the GR subzone suggests that, in contrast to the AC subzone, some diﬀerentiation has taken place in the formation of this subzone (Fig. 9). Compositionally-uniform cores indicate that plagioclase crystallised without signiﬁcant fractionation, which suggests that the eﬀective size of the melt reservoir in equilibrium with the plagioclase was large enough and that the melt was thoroughly stirred. The vertical compositional gradient in the GR subzone, therefore, is interpreted to be a record of the melt composition evolved as the GR subzone grew (i.e., the upward migration of the boundary layer). It is conceivable, therefore, that the GR subzone represents successively frozen magmatic boundary layers that have developed above the AC subzone. It was concluded from the observation of crystal grain size that olivine has grown signiﬁcantly in the formation of the GR subzone (Hoshide et al. 2006a). Apparently, plagioclase has grown considerably as well. It was observed that, in the Lower Zone, olivine crystals are often partly included in the mantle of the plagioclase and are in contact with the outlines of the calcic cores (Figs 4c, 5a, 6a). It is plausible, therefore, that the calcic plagioclase and olivine came closer during the crystal settling and ‘collided’

with each other (i.e., crystal impingement) before the plagioclase overgrowth, forming a crystal-supported ‘framework structure’. 4.1.2. Resorption of calcic plagioclase and the formation of the high-An edges. The presence of hydrous phases, such as amphibole and chlorite at the embayments of the calcic cores, indicates that the resorption and probably the formation of high-An edges may have occurred in a hydrous environment that was created by the introduction (or ﬂuxing) of H2O. The resorption may occur when melt becomes undersaturated with plagioclase, and such a disequilibrium situation may be realised if H2O is introduced in the system as follows. An introduction of H2O would depress the liquidi (and solidi) of anhydrous phases such as plagioclase, olivine and cpx. Because the extent of the depression is greater for plagioclase than for olivine and cpx (e.g., McBirney 1987), the plagioclase ﬁeld shrinks, and therefore, the liquid originally coexisting with olivine and plagioclase becomes undersaturated with these silicates and, therefore, plagioclase (and olivine as well) simply starts to dissolve (Fig. 10, the ‘simple dissolution’ of Tsuchiyama 1985). The source of H2O may be sought in the lower part of the boundary layer, because the H2O content in the melt increases rapidly as fractionation advances. More advanced resorption observed in the anorthosite implies more pervasive and extensive introduction of H2O there than in the host. Following McBirney (1987), the present authors interpret the formation of high-An edges in the context of the hydrous crystallisation as follows. Suppose liquid L1 and plagioclase S1 once equilibrated (Fig. 10). When H2O comes in, both liquidus and solidus are depressed according to the introduced H2O

246

TAKASHI HOSHIDE AND MASAAKI OBATA

Figure 10 (a) Isothermal sections of a hypothetical liquidus diagram of an olivine–Ab–An-(+H2O) system. Dry (black) and wet (red) isothermal sections at the same temperature and the same total pressure but of diﬀerent activities of H2O are drawn for comparison. Each ﬁeld is labelled for wet condition for clarity only. Projections of Ol–Pl cotectic lines are partly drawn as dashed lines. When H2O is introduced in the dry system in which liquid L1 and plagioclase S1 were equilibrated, the original liquid L1 becomes undersaturated with plagioclase (and olivine) and the plagioclase (S1) starts to dissolve, forming a thin ﬁlm of hydrous liquid L2 on the surface of the dissolving plagioclase crystals. L2–S2 is an equilibrium tie line between the hydrated liquid and plagioclase. (b) Schematic projections of the plagioclase liquidus and solidus from H2O and olivine apexes onto the plagioclase-temperature plane. Dry (black) and wet (red) projections are compared. L1, L2, S1, S2 are projections of those in the isothermal sections in (a). See text for more explanation.

content, and then the original liquid becomes undersaturated and so the plagioclase (S1) starts to dissolve, forming a thin ﬁlm of hydrous liquid L2 on the surface of the dissolving plagioclase crystals. The formation of liquid L2 makes a steep compositional gradient from the original composition L1, away from the plagioclase, toward L2 on the plagioclase surface. If the ambient temperature decreases faster than the elimination of this compositional gradient by diﬀusion in the melt, the new plagioclase that is precipitated on the dissolved plagioclase surface will be strongly zoned, reﬂecting this compositional gradient as follows. The ﬁrst plagioclase to precipitate will be very close to S2, which is in equilibrium with L2 and is more An-rich than the original composition S1. Upon

cooling, a sharp zoning will be formed from this S2 toward composition S1 which is in equilibrium with the original liquid L1, thereby forming the high-An edge. After that, normal zoning due to ordinary fractional crystallisation follows, which should show a gentler compositional slope than that in the high-An edge. Dissolution of olivine as well as of plagioclase should occur when H2O comes in and, because of the shift of the olivine– plagioclase cotectic line toward plagioclase, the original liquid enters into the olivine ﬁeld (Fig. 10a) and then, upon cooling, olivine starts to grow before the liquid gets saturated with plagioclase again. It is probable that hydrous melts, which are now represented by amphibole clots, were those entrapped in these growing olivines. Similar amphibole-bearing clots have been observed in olivine, apatite and chromite crystals in many plutonic rocks (Boudreau et al. 1986; Boudreau 1999; Jakobsen et al. 2005; Li et al. 2005). Jakobsen et al. (2005) interpreted those amphibole clots in olivine from the Skaergaard Intrusion to represent trapped immiscible liquids at some early stages of fractionation. The similar occurrences and mineralogies of these amphibole clots may suggest that similar hydrous processes as advocated above for the Murotomisaki Gabbroic Intrusion (MGI) were operative in many plutonic environments. Pringle et al. (1974) observed similar calcic edges of plagioclase in a diabase sheet from Grand Manan Island (Canada) and ascribed it to a large pressure drop when host magma ascended from a deep magma reservoir. However, such a hypothesis cannot be applied to the MGI, because the calcic cores of plagioclase in the MGI are generally larger than the plagioclase phenocrysts from the chilled margin and, therefore, the formation of the high-An edges must be after the emplacement of the magma. Certain pressure drop may occur in a magma reservoir if eruption occurs in overpressurised circumstances (e.g., Williams & McBirney 1979). The observed compositional ‘jump’ of the high-An edge (about 10 An%) corresponds to as much as 5 kbar pressure drop (Takagi et al. 2005). This is too large to be accounted for by volcanic eruptions (Williams & McBirney 1979). Similar high-An edges of plagioclase have also been described from the Skaergaard Intrusion by Maaløe (1976), who ascribed the feature to a disequilibrium overgrowth of plagioclase from supercooled magma. The supersaturation hypothesis has been employed to explain the oscillatory zoning of plagioclase (e.g., Sibley et al. 1976), but will not ﬁt the explanation for such a sharp and single An-edge as observed in the MGI. An alternative hypothesis has therefore been presented for the MGI for the formation of high-An edges, combining the resorption by means of an introduction of H2O (H2O ﬂuxing) from the lower horizons of the boundary layer. 4.1.3. Origin of the sodic mantle. Overgrowth of the sodic mantle occurred after the resorption events described above. As emphasised above, the important diﬀerence between the host gabbro and the anorthosites in the GR subzone lies in the composition of the overgrown mantle of the plagioclase (Fig. 7a, b). Whilst the An-content gradually decreases from the core to the mantle of plagioclase in the host gabbro (Type A), it abruptly drops in the mantle in the anorthosites, marking a clear compositional gap at the edge–mantle boundary for the latter (Type B). The compositional gap in the plagioclase in the anorthosites may be ascribed to a discontinuous change in composition of the melt surrounding the plagioclase crystals. Such a change of melt composition cannot be explained by the introduction of H2O alone, as advocated above, and an additional introduction of fractionated melt components would be required. The introduction of the hypothetical fractionated melt was probably restricted to the anorthosite and

ZONING OF PLAGIOCLASE IN A LAYERED INTRUSION

247

Figure 11 (a) Schematic picture of a stratiﬁed basal boundary layer. The boundary layer is divided into three zones: Zone 1, plagioclase and olivine precipitate; Zone 2, hydrous ﬂux melting and melt segregation (podiform gabbroic pegmatites and anorthosite layers); Zone 3, source of H2O and fractionated melts. See text for explanation. Thermal and H2O concentration gradients are only schematic. Fractionated melt reaches H2O saturation in Zone 3. (b) Advancing two boundary layers produce a stratiﬁed layered gabbro.

gabbro pegmatites. In other words, the anorthosite and the gabbro pegmatites are the very sites that were preferentially attacked by the ﬂux of fractionated melts accompanied by H2O. The gentle An-decrease zoning in the mantle reﬂects subsequent fractional crystallisation that took place in the interstitial melts. The source of this hypothetical fractionated melt is also sought in the lower part of the boundary layer (Fig. 11). It is conceivable that both H2O and fractionated melt come together from the same lower horizons. Fractionated melt will be enriched with H2O, but considering the probable H2Osaturated situation, it is likely that H2O-rich ﬂuids and the fractionated melt came together. Plagioclase crystals from the picritic ﬂoor of the podiform gabbroic pegmatite hardly contain these calcic cores (Figs 6c, 7c). Their core composition is comparable to the mantle of the plagioclase from the anorthositic roof (Fig. 9). It is conceivable, therefore, that the crystallisation of plagioclase in the picritic ﬂoors took place concomitantly, maintaining chemical equilibrium with the mantle overgrowth of the plagioclase in the anorthositic roofs, i.e., after the invasion (or inﬁltration) of the fractionated melt. Similar to the plagioclase core in the anorthositic roofs, the olivine in the picritic ﬂoors has nearly the same composition as those in the adjacent host olivine gabbros. Therefore, it is conceivable that both the calcic plagioclase and the olivine formed together in the host gabbro and that they were separated and segregated when the fractionated melt came in.

4.2. Middle Zone The Middle-Zone coarse gabbros contain three types of zoning, B, B# and C (Fig. 8, Table 1). Many plagioclases from the Middle Zone are more sodic than those from the Upper and Lower Zones (Type-C), while those in some samples (at the 152- and 153-m levels) are strongly calcic and have a homogeneous calcic core (Types B and B#). The Type C plagioclase may represents in-situ crystallisation products from the ambient fractionated melt. Type B and B# plagioclases on the other hand must be exotic in origin. Because plagioclase with the high-An edges (i.e. Types A or B) cannot be found in the Upper Zone, the source of these plagioclases must be sought in

the Lower Zone. Occurrence of the Type-B and B# plagioclase in the Middle Zone may be understood by considering the diapirism of anorthositic crystal mushes generated from the GR subzone below, as advocated by Hoshide et al. (2006b). In the samples containing Type-B and B# plagioclase, the Ancontents of plagioclase cores have a wide range amongst the diﬀerent grains within one sample. This disequilibrium amongst the diﬀerent grains suggests repeated supplies of diapiric anorthositic crystal mushes from the advancing basal boundary layer and incomplete magma mixing.

4.3. Upper Zone The Upper Zone contains two types of zoning, B# and C (Fig. 8 and Table 1). In the Upper Zone, the maximum value of An-content of the plagioclase core (Fig. 9) has an apparently vertical gradient. Some of the plagioclases forming the vertical compositional gradient have compositional gaps between the cores and the mantles (i.e., Type-B#). The compositional gaps may suggest mixing between the crystal and the fractionated melt, as inferred for the Lower Zone. It is possible that the Type B# plagioclase has its origin in the Lower Zone as in the Middle Zone, but full understanding of the origin and the processes of the Upper Zone must await further, more detailed, investigations of this zone.

5. Segregation and diapirism of the anorthositic crystal mushes Another important eﬀect of H2O is lowering the viscosity of silicate melt. The H2O-charged ﬂux-melting zone within the boundary layer may become ﬂuid-dynamically unstable, because H2O eﬀectively reduces the viscosity of silicate melt and thus increases the readiness of segregation in crystal mushes (McBirney 1987), causing an eﬀective separation of lighter plagioclase crystals from the heavier olivine forming a modally graded layering (Fig. 2c), as observed in the base of the GR subzone. The podiform gabbroic pegmatite and the anorthosite may represent evolved forms of such a modallylayered structure. If the H2O-enriched magma is segregated, forming a pod or layer, it will gain more buoyancy because the

248

TAKASHI HOSHIDE AND MASAAKI OBATA

buoyancy is proportional to the volume of the masses, which would cause a gravitational instability. The wavy structure of the anorthosite roof and the upper surface of the anorthosite layer may be a manifestation of such instability. As crystallisation advances within the podiform gabbroic pegmatite, residual melt will become even more hydrous. This is consistent with the formation of pegmatitic ﬁlling between the anorthositic roof and the picritic ﬂoor, and also accounts for the observed coarsening of olivine crystals in the GR subzone. The occurrence of such buoyant hydrous melts in the anorthosite will enhance the buoyancy of the anorthositic magma and help the diapiric plume to develop. These plumes will eventually detach from their source, leaving the podiform gabbroic pegmatite with little anorthositic roof, as observed. The fact that the proportion of the anorthosite roof is very small in some podiform pegmatites (Fig. 3d) may indicate that substantial amounts of anorthositic magmas may have already escaped in the form of plumes from the site of their generation. The quantitative aspect of this process, i.e., the amount of the plume generated and the extent of its eﬀect on the magmatic evolution, must be examined by studying mass balance using more extensive whole-rock data, which is a subject of future work.

The Upper Zone may have been formed at the same time by another moving boundary layer descending from the top of the magma body (Fig. 11b). However, the detailed processes in this descending boundary layer have not been discussed. The Middle Zone may represent the last layer of freezing sandwiched by these two moving boundary layers. More detailed analysis of the Middle and Upper zones is necessary for a full understanding of the whole process of magmatic diﬀerentiation.

7. Acknowledgements We are grateful to Dirk Spengler and Tetsuo Kawakami of Kyoto University for their discussions and critical reading of the manuscript and to A. Toramaru (Kyushu University) for valuable discussions in the ﬁeld. We also acknowledge T. Akatsuka for his permission for the use of unpublished data, and to H. Sato and U. Honma for arranging the electron microprobe facility at Venture Business Laboratory of Kobe University. The paper has beneﬁted from critical and constructive criticisms of J. H. Be´dard and R. A. Wiebe. This study was ﬁnancially supported by a Grant-in-Aid for Research Fellow of the Japan Society for the Promotion of Science (No.19·4363) to TH.

6. Summary and synthesis The processes deciphered from the analysis of the zonal structure of the MGI plagioclase may be summarised in the schematic picture in Figure 11. The whole process of the formation of the GR subzone (in the Lower Zone) may be understood in terms of a moving boundary layer in a crystallising magma body. The boundary layer, which contains steep thermal and chemical gradients, should be internally stratiﬁed in a steady state in three zones, 1, 2 and 3, as in Figure 11a. Zone 1, just beneath the crystallisation front, is that of crystal growth and crystal settling. Olivine and plagioclase steadily grow as the cooling advances and grown crystals will start to settle, probably leading to the ‘crystal impingement’. Zone 2 is where H2O-ﬂuxed remelting and local crystal-melt segregation occurs, caused by the introduction of H2O and fractionated melts ascending from Zone 3 below. The modal layering, podiform gabbroic pegmatites and anorthositic layers were probably formed in this zone. Microscopically high-An edges formed around the resorbed plagioclase crystals upon further cooling. Zone 3 is the source of the H2O and hydrous fractionated melts which should be saturated with H2O. Mixtures of H2O-rich bubbles and fractionated melts probably rise together from this zone and get into Zone 2. Formation of the plagioclase sodic mantle and precipitation of augite occur in this zone. The compositional layering developed in the second layer will be frozen in this layer and, as the boundary layer migrates upward, accrete to the growing GR subzone just behind the moving boundary layer. Zone 2 is also a source of anorthositic crystal mush plumes, which may feed the main magma body above the boundary layer. The possibility that plagioclase crystals with high calcic cores (i.e., Type B and Type B#) found in the Middle Zones were derived from these anorthositic plumes was discussed previously. This is a similar process to the ‘constitutional zone reﬁning’ proposed by McBirney (1987) and Boudreau (1999), except for the diapiric ascent of anorthositic crystal mushes. H2O accumulation in Zone 2 is responsible for the segregation and formation of the compositional layering as originally proposed by McBirney (1987). Quantitative aspects of the eﬀects of the ascending plumes generated from Zone 2 remains to be examined with more geochemical data.

8. References Akatsuka, T., Obata, M. & Yokose, H. 1999. Formation of layered structure in the Murotomisaki gabbroic complex, especially picrite gabbro, Kochi Prefecture, Japan–Quantitative evaluation of crystal accumulation. Journal of the Geological Society of Japan 105 (11), 771–88. [In Japanese with English abstract.] Boudreau, A. E. 1999. Fluid Fluxing of Cumulates: the J-M Reef and Associated Rocks of the Stillwater Complex, Montana. Journal of Petrology 40, 755–72. Boudreau, A. E., Mathez, E. A. & Mccallum, I. S. 1986. Halogen Geochemistry of the Stillwater and Bushveld Complexes: Evidence for Transport of the Platinum-Group Elements by Cl-Rich Fluids. Journal of Petrology 27, 967–86. Brandeis, G. & Jaupart, C. 1986. On the interaction between convection and crystallization in cooling magma chambers. Earth and Planetary Science Letters 77, 345–61. Costa, A., Blake, S & Self, S. 2006. Segregation processes in vesiculating crystallizing magmas. Journal of Volcanology and Geothermal Research 153, 287–300. Hamamoto, R. & Sakai, H. 1987. Rb–Sr age of granophyre associated with the Muroto-misaki gabbroic complex. Science Reports of the Faculty of Science, Kyushu University, Geology 15 (1), 131–5. [In Japanese with English abstract.] Hoshide, T., Obata, M. & Akatsuka, T. 2006a. Crystal settling and crystal growth of olivine in the magmatic diﬀerentiation – the Murotomisaki Gabbroic Complex, Shikoku, Japan –. Journal of Mineralogical and Petrological Sciences 101, 223–39. Hoshide, T., Obata, M. & Akatsuka, T. 2006b. Magmatic diﬀerentiation by means of segregation and diapiric ascent of anorthositic crystal mush – the Murotomisaki Gabbroic Complex, Shikoku, Japan. Journal of Mineralogical and Petrological Sciences 101, 334–9. Jakobsen. J. K., Veksler, I. V., Tegner, C. & Brooks, C. K. 2005. Immiscible iron- and silica-rich melts in basalt petrogenesis documented in the Skaergaard intrusion. Geology 33, 885–8. Kodama, K., Taira, A., Okamura, M. & Saito, Y. 1983. Paleomagnetism of the Shimanto Belt in Shikoku, southwest Japan. In Hashimoto, M. & Uyeda, S. (eds) Accretion Tectonics in the Circum-Pacific Regions, 231–41. Tokyo: Terra Scientiﬁc Publishing Company. Kodama, K. & Koyano, T. 2003. Paleomagnetism and magnetic fabric of a maﬁc intrusive body and surrounding sedimentary rocks in the shimanto belt, southwest Japan: structural evolution of an accretionary prism recorded in magnetization of fore-arc intrusive rocks. IUGG 2003 Scientific Program Abstract, GAI.06. Sapporo, Japan: International Union of Geodesy and Geophysics. Li, C., Ripley, E. M., Sarkar, A., Shin, D. & Maier, W. D. 2005. Origin of phlogopite-orthopyroxene inclusions in chromites from

ZONING OF PLAGIOCLASE IN A LAYERED INTRUSION the Merensky Reef of the Bushveld Complex, South Africa. Contributions to Mineralogy and Petrology 150, 119–30. Maaløe, S. 1976. The zoned plagioclase of the Skaergaard Intrusion, East Greenland. Journal of Petrology 17, 398–419. Mangan, M. & Marsh, B. D. 1992. Solidiﬁcation front fractionation in phenocryst-free sheet-like magma bodies. Journal of Geology 100, 605–20. Marsh, B. D. 1988. Crystal capture, sorting, and retention in convecting magma. Geological Society of America Bulletin 100, 1720–37. McBirney, A. R. 1987. Constitutional zone reﬁning of layered maﬁc intrusions In Parsons, I. (ed) Origins of Igneous Layering, 437–52. Dordrecht, Holland: D. Reidel Publishing. Pringle, G. J., Trembath, L. T. & Pajari, G. E. 1974. Crystallization history of a zoned plagioclase. Mineralogical Magazine 39, 867– 77. Sibley, D. F., Vogel, T. A., Walker, B. M. & Byerly, G. R. 1976. The origin of oscillatory zoning in plagioclase; a diﬀusion and growth controlled model. American Journal of Science 276, 275–84.

249

Tait, S. & Jaupart, C. 1992. Compositional convection in a reactive crystalline mush and melt diﬀerentiation. Journal of Geophysical Research 97, 6735–56. Takagi, D., Sato, H. & Nakagawa, M. 2005. Experimental study of a low-alkali tholeiite at 1–5 kbar: optimal condition for the crystallization or high-An plagioclase in hydrous arc tholeiite. Contributions to Mineralogy and Petrology 149, 527–40. Tsuchiyama, A. 1985. Dissolution kinetics of plagioclase in the melt of the system diopside-albite-anorthite, and origin of dusty plagioclase in andesites. Contributions to Mineralogy and Petrology 89, 1–16. Turner, J. S., Huppert, H. E. & Sparks, R. S. J. 1986. Komatiites II: Experimental and theoretical investigations of post-emplacement cooling and crystallization. Journal of Petrology 27, 397–437. Williams, H. & McBirney, A. R. 1979. Vulcanology. San Francisco, California: Freeman, Cooper. 397 pp. Yajima, T. 1972. Petrology of the Murotomisaki gabbroic complex. The Journal of the Japanese Association of Mineralogists, Petrologists and Economic Geologists 67, 218–41.

MS received 30 November 2007. Accepted for publication 16 May 2008 (Stellenbosch); 15 January 2009 (RSE).

ISBN 978-0-8137-2472-0

THE GEOLOtiCAL SOCIE OF AMERICA®

(} 0

3300 Penrose Place • P.O. ~Hti~ii~"·' 2i' Boulder, CO 803~1-9140,

9 780813 724720

Frontiers of Fundamental Physics : Proceedings of the Sixth International Symposium

Geologic Evolution of the Barberton Greenstone Belt, South Africa (GSA Special Paper 329)

IUTAM Symposium on Theoretical, Computational and Modelling Aspects of Inelastic Media: Proceedings of the IUTAM Symposium Held at Cape Town, South Africa, January 14-18, 2008

The Kleene Symposium: Proceedings of the Symposium Held June 18-24, 1978 at Madison, Wisconsin, U.S.A

IUTAM symposium on vibration control of nonlinear mechanisms and structures: proceedings of the IUTAM symposium held in Munich, Germany, 18-22 July 2005

Sixth Hutton Symposium on the origin of granites and related rocks: proceedings of a symposium held in Stellenbosch, South Africa, 2-6 July 2007 (GSA Special Paper 472)

Frontiers of Fundamental Physics : Proceedings of the Sixth International Symposium

Geologic Evolution of the Barberton Greenstone Belt, South Africa (GSA Special Paper 329)

IUTAM Symposium on Theoretical, Computational and Modelling Aspects of Inelastic Media: Proceedings of the IUTAM Symposium Held at Cape Town, South Africa, January 14-18, 2008

The Kleene Symposium: Proceedings of the Symposium Held June 18-24, 1978 at Madison, Wisconsin, U.S.A

Electromagnetic Materials: Proceedings of the Symposium P, ICMAT 2007

Towards Next Generation Grids: Proceedings of the CoreGRID Symposium 2007

Proceedings of the symposium on algebraic geometry in East Asia

Electromagnetic Materials: Proceedings of the Symposium P, ICMAT 2007

IUTAM symposium on vibration control of nonlinear mechanisms and structures: proceedings of the IUTAM symposium held in Munich, Germany, 18-22 July 2005

Proceedings of the Symposium on Differential Equations and Dynamical Systems

Patras Logic Symposium: Proceedings

Proceedings of the Third Scandinavian Logic Symposium

Posture, Locomotion, and Paleoecology of Pterosaurs (GSA Special Paper 376)

Construction and Applications of Conformal Maps, Proceedings of a Symposium

Proceedings of the Second Scandinavian Logic Symposium

L.E.J.Brouwer Centenary Symposium Proceedings

Construction and Applications of Conformal Maps, Proceedings of a Symposium

Ancient Seismites (GSA Special Paper 359)

Coal Systems Analysis (GSA Special Paper 387)

IUTAM Symposium on Unsteady Separated Flows and their Control: Proceedings of the IUTAM Symposium (IUTAM Bookseries)

Proceedings of the Third Japan-USSR Symposium on Probability Theory

Selected proceedings of the Symposium on Estimating Functions

Aircraft Loading Action Problems (Proceedings of a Symposium held at Farnborough on 28th October, 1966)

Proceedings of the Thirteenth Symposium on Naval Hydrodynamics

Proceedings of the Second Japan-USSR Symposium on Probability Theory

Proceedings of the Navy Symposium on Aeroballistics (10th)

Proceedings of the Ninth International Symposium on Cyclodextrins

Volcanic Rifted Margins (GSA Special Paper 362)

Selected Proceedings of the Symposium on Inference for Stochastic Processes

Proceedings of Liverpool Singularities Symposium II

Sixth Hutton Symposium on the origin of granites and related rocks: proceedings of a symposium held in Stellenbosch, South Africa, 2-6 July 2007 (GSA Special Paper 472)

Frontiers of Fundamental Physics : Proceedings of the Sixth International Symposium

Geologic Evolution of the Barberton Greenstone Belt, South Africa (GSA Special Paper 329)

IUTAM Symposium on Theoretical, Computational and Modelling Aspects of Inelastic Media: Proceedings of the IUTAM Symposium Held at Cape Town, South Africa, January 14-18, 2008

The Kleene Symposium: Proceedings of the Symposium Held June 18-24, 1978 at Madison, Wisconsin, U.S.A

Electromagnetic Materials: Proceedings of the Symposium P, ICMAT 2007

Towards Next Generation Grids: Proceedings of the CoreGRID Symposium 2007

Proceedings of the symposium on algebraic geometry in East Asia

Electromagnetic Materials: Proceedings of the Symposium P, ICMAT 2007

IUTAM symposium on vibration control of nonlinear mechanisms and structures: proceedings of the IUTAM symposium held in Munich, Germany, 18-22 July 2005

Proceedings of the Symposium on Differential Equations and Dynamical Systems

Patras Logic Symposium: Proceedings

Proceedings of the Third Scandinavian Logic Symposium

Posture, Locomotion, and Paleoecology of Pterosaurs (GSA Special Paper 376)

Construction and Applications of Conformal Maps, Proceedings of a Symposium

Proceedings of the Second Scandinavian Logic Symposium

L.E.J.Brouwer Centenary Symposium Proceedings

Construction and Applications of Conformal Maps, Proceedings of a Symposium

Ancient Seismites (GSA Special Paper 359)

Coal Systems Analysis (GSA Special Paper 387)

IUTAM Symposium on Unsteady Separated Flows and their Control: Proceedings of the IUTAM Symposium (IUTAM Bookseries)

Proceedings of the Third Japan-USSR Symposium on Probability Theory

Selected proceedings of the Symposium on Estimating Functions

Aircraft Loading Action Problems (Proceedings of a Symposium held at Farnborough on 28th October, 1966)

Proceedings of the Thirteenth Symposium on Naval Hydrodynamics

Proceedings of the Second Japan-USSR Symposium on Probability Theory

Proceedings of the Navy Symposium on Aeroballistics (10th)

Proceedings of the Ninth International Symposium on Cyclodextrins

Volcanic Rifted Margins (GSA Special Paper 362)

Selected Proceedings of the Symposium on Inference for Stochastic Processes

Proceedings of Liverpool Singularities Symposium II

Recommend Documents